Podcasts about Apis

What happens when an organization writes careful AI governance policies but its infrastructure cannot enforce any of them? In this episode of Tech Talks Daily, I speak with Sabina Anja, Chief Technologist at Broadcom within the VMware Cloud Foundation division, about the infrastructure controls required as AI agents move from generating answers to accessing data, calling APIs, modifying systems, and triggering work. Sabina brings experience from both sides of enterprise technology. She remembers cabling networks, dealing with unstable infrastructure, and receiving those weekend calls when downtime had already upset the business. That background informs her belief that ambitious AI programs cannot succeed without stable, observable, and enforceable infrastructure beneath them. Many organizations are repeating a familiar pattern. Business teams adopt AI services before IT has established visibility, ownership, or control. The terminology may have changed from shadow IT to shadow AI, but the management problem remains. Sabina argues that CIOs first need an inventory of agents, nonhuman identities, data access, processes, and accountable owners. The risk becomes greater because agents behave differently from people. They operate across multiple systems at machine speed and can perform repeated actions without appreciating the wider business outcome. An agent does not need malicious intent to cause disruption. Excessive permissions, flat networks, inconsistent access rules, and years of deferred infrastructure work can give it plenty of opportunities. Sabina recommends brokered access rather than direct access, alongside dedicated virtual machines or namespaces, microsegmentation, lateral security, east-west policy controls, and tamper-evident logging. Organizations also need to define which data an agent can view, modify, or move, especially when sovereignty and regulatory requirements apply. One of Sabina's most memorable ideas is to treat an AI agent like a superhuman contractor. It should have a defined purpose, a named manager, a clear access specification, an activity record, and an end date. Additional permissions should be earned through evidence of reliable behavior rather than granted on the first day. She also warns about agent debt. AI systems are developing rapidly, so an agent created today may become outdated within months. Sabina recommends assuming that many agents will expire after six to nine months rather than allowing forgotten systems and permissions to accumulate indefinitely. For CIOs wanting an immediate test, her advice is straightforward. Create an inventory of nonhuman identities with production access. Then select one agent and examine every part of the infrastructure it attempted to reach. The question is not simply whether the application produced the expected result. Leaders should ask whether the agent entered systems, networks, or data stores that nobody expected it to access. We also challenge the familiar claim that AI agents will take everybody's jobs. Sabina sees an opportunity to remove repetitive tasks and give technology professionals new skills, although she warns that agents may behave like teenagers armed with infrastructure permissions. They may not take your job, but they could become remarkably good at testing your patience. I'd love to hear your thoughts. Does your organization know how many AI agents have production access and who is accountable for each one?  

In this episode of Two Bees in a Podcast, Amy Vu and Dr. Jamie Ellis discuss Apis mellifera mellifera with Dr. Mark Barnett, researcher at the Roslin Institute at the University of Edinburgh in Scotland and co-founder of Beebytes. Check out our website: www.ufhoneybee.com for additional resources from today's episode. 

Topics covered in this episode: Web UIs for your reverse proxy Wagtail 8.0 is hot off the presses RISC-V is now officially supported by CPython Django's annual releases make every version an LTS Extras Joke Watch on YouTube About the show Sponsored by Logfire from Pydantic: pythonbytes.fm/logfire Connect with the hosts Michael: Mastodon / BlueSky / X / LinkedIn Calvin: Mastodon / BlueSky / X / LinkedIn Show: Mastodon / BlueSky / X Join us on YouTube at pythonbytes.fm/live to be part of the audience. Usually Tuesday at 7am PT. Older video versions available there too. Finally, if you want an artisanal, hand-crafted digest of every week of the show notes in email form? Add your name and email to our friends of the show list, we'll never share it. Michael #1: Web UIs for your reverse proxy Traefik, nginx, and Caddy all sit in front of a lot of self-hosted infrastructure, and all three are configured by hand-editing files. Three active projects put a control plane on top: Traefik Manager (Python + Flask), Nginx UI (Go + Vue), and caddy/ui (React + Node). All three are additive rather than replacements - none of them take ownership of your config away from you - which is the part that matters when the thing has write access to production routing. Traefik Manager is the Python one: Flask 3.1 and Gunicorn for the control plane, a lightweight Go agent for remote instances, currently v1.10.0 with an Android companion app. Nginx UI is a single Go binary at 11.3k stars, with a block-style config editor, an Ace editor doing LLM completion on nginx syntax, and an MCP server so agents can drive it. caddy/ui runs as two containers next to your existing Caddy, reads and writes your Caddyfile directly, and uses Caddy's /adapt API to validate before reload - no Docker socket required. Each one edits the config the underlying server already reads, so your files stay the source of truth and you can drop the UI without unwinding anything. Undo is a first-class feature across all three - timestamped backups with optional Git history, config version compare and restore, Caddyfile snapshots with one-click rollback. Observability is where they diverge: Traefik Manager does CrowdSec and a visual route map, Nginx UI does server metrics, caddy/ui streams access logs over SSE and pulls p50/p95/p99 off Caddy's Prometheus endpoint. Maturity spread is wide - Nginx UI has 11.3k stars, caddy/ui has 4 and was built in a single Claude session - and caddy/ui ships with auth off by default, so set CADDY_UI_USER and JWT_SECRET before it goes anywhere near a public interface. Calvin #2: Wagtail 8.0 is hot off the presses Link: https://github.com/wagtail/wagtail/releases/tag/v8.0 Custom base page models are now supported, so projects aren't locked into subclassing Wagtail's Page as shipped (Matt Westcott). New v3 REST API handles both read and write CMS operations, a first for Wagtail's API. A global registry for permission policies, plus full customizability for the remaining page views via PageViewSet. AVIF and WebP images are no longer auto-converted to PNG by default, a real behavior change to watch on upgrade. Five security fixes: page admin API restrictions, document identification by SHA1 hash, descendant collections in the Documents/Images API, snippet copy permissions, and the page translation endpoint. Formalized Django 6.1 support, and CI now runs on uv with a lockfile. Sponsor: Logfire from Pydantic Your AI agent failed at 2am. Was it the model? A tool call? The database? Most observability tools can't tell you, because they only see part of your stack. Pydantic Logfire sees all of it. One trace across your agents, LLMs, APIs, and database. Down to the infrastructure: services, Kubernetes, and hosts. It's built on OpenTelemetry, with SDKs for Python, TypeScript, and Rust, and it works with any OTel-compatible language. Every prompt, token count, and cost, right next to your vector searches and API calls. You query everything with Postgres-compatible SQL. And so can your coding agent, through the Logfire MCP server. Stop guessing. Read the trace. Pydantic Logfire. AI, it's still just engineering. Visit pythonbytes.fm/logfire today and sign up today. Get 10M records free every month, no card required. You can even click “Onboard with your coding agent” to copy a prompt to have claude or codex integrate Logfire into your app. Thanks to Pydantic for supporting the show. Calvin #3: RISC-V is now officially supported by CPython Link: https://blog.python.org/2026/08/riscv-now-officially-supported/ CPython added RISC-V as a tier 3 platform under PEP 11, specifically the 64-bit Linux target riscv64-unknown-linux-gnu. RISC-V is an open ISA anyone can implement, unlike x86 and ARM, and its market is projected to quadruple by 2032. The RISE Project donated real RISC-V machines for buildbots; the author's work was funded by a Sovereign Tech Agency fellowship. What changes: the port is now a maintained compatibility target, so CPython changes are less likely to quietly break it. What doesn't: no python.org installers, no binary wheel parity for native extensions. Next up: RISC-V runners in CPython CI for pre-merge feedback, then a push toward tier 2, plus architecture-specific optimizations. The ask is testing. If you have RISC-V hardware, build CPython, run your test suite, file what breaks. Tier 3 is the weakest support tier. PEP 11 tier 3 requires a core developer contact and a buildbot, but failures on tier 3 platforms explicitly do not block a release. Saying "ongoing CI/testing expectations" oversells it. The honest bit is "someone is now on the hook for it, and breakage gets noticed," not "it's guaranteed working." Worth the caveat that this is Linux SBCs, not microcontrollers. A VisionFive 2 counts, an ESP32-C6 or Pico 2 does not. Those are 32-bit non-Linux parts where MicroPython is still the answer. Michael #4: Django's annual releases make every version an LTS Starting with Django 2028, Django will move to one January feature release per year, adopt calendar-based version numbers, and support every release for three years. The old distinction between standard and LTS releases disappears, giving teams a predictable annual upgrade path that aligns more closely with Python's own release and support cadence. Every Django release becomes the safe, long-supported choice, so teams no longer need to wait for a specially designated LTS version or absorb two years of changes at once. Each release gets one year of mainstream bug fixes followed by two years of security and data-loss fixes. New releases support the three latest Python versions and add the next Python release during their first year. Calendar versioning begins with Django 2028, followed by Django 2029 and so on. Three Django versions will be supported at any time, giving third-party packages a clearer rolling target. Nothing changes before 2028, and existing commitments for Django 5.2 LTS and 6.2 LTS remain in place. Extras Calvin: The Python docs now document the time complexity of built-in types https://docs.python.org/3.16/library/time-complexity.html Thinking in Python - Bruce Eckel's free book https://thinkinginpython.com/ Michael: prune_uv_pythons.py - Prune uv-managed Python installs, keeping only the newest patch per minor version Runs automatically in my system “upgrade” script: upgrade-output-2026.png Started using Ollama cloud models for my Hermes assistant. Thanks to Jeff Triplett I learned they are not just local models. Joke: The Tao of Programming - Book Seven: Corporate Wisdom

DESCRIPTION Welcome to The Ecommerce Braintrust podcast, brought to you by Julie Spear, Head of Retail Marketplace Services, and Jordan Ripley, Director of Retail Account Management. Today, we're tackling a topic that has evolved from a conference panel talking point into one of the most significant media budget shifts in digital advertising: Amazon DSP for non-endemic brands.  Historically, brands that don't actually sell physical products on Amazon's marketplace have questioned why they should buy Amazon media. Today the conversations we're having with brands are very different.  Joining us today to unpack why 2026 is the turning point for non-endemic media buying is our very own Director of Retail Media at Acadia, Ross Walker.  Let's dive in!    Quote: If you're a non-endemic brand and you're hitting a point of consistent diminishing returns with your traditional media channels, that's the first signal that you should consider testing into a new media channel like ADSP. Ross Walker   KEY TAKEAWAYS In this episode, Julie, Jordan, and Ross discuss: Endemic vs. non-endemic, simplified: It's really just "sells on Amazon" vs. "doesn't sell on Amazon" - the jargon is mostly industry inertia. Amazon DSP has hit parity with The Trade Desk: Years of building out premium CTV/STV, audio, digital out-of-home, and owned-channel inventory means Amazon can now support the same kind of full-funnel programmatic buys as legacy DSPs. The real edge is commerce data, at a lower cost: No other DSP has Amazon's depth of purchase-intent signal, and Amazon has intentionally kept platform and audience fees low to win market share while it can. Premium CTV reach is a genuine unlock: ADSP now reaches roughly 80 million authenticated CTV households a month through partners like Netflix, Disney+, and Roku - inventory that used to be exclusive to players like The Trade Desk. Attribution is no longer the weak link: Tools like Amazon Marketing Cloud, the Amazon Ad Tag, and conversion APIs let non-endemic brands tie a Netflix or Roku impression directly to a D2C sale, lead form, or sign-up - solving a measurement gap that's plagued programmatic for years. Endemic brands get more flexibility too: Brands that do sell on Amazon can now send DSP traffic to their own site or other retailers (Walmart, Ulta, Home Depot), though the richest behavioral/custom audiences are still reserved for traffic that stays on Amazon. Signals to know it's time to test ADSP: Diminishing returns on Meta/Google/Trade Desk, or an audience with strong, specific purchase intent are both green lights to start testing. What still differentiates DSPs going forward: As capabilities converge across platforms, the real differentiator becomes each platform's unique owned audiences and media properties (e.g., Amazon's Prime Video, Walmart's owned data via The Trade Desk).

Sathiesh Veera, a GenAI Solutions Architect at At&T, speaks with host Brijesh Ammanath about the data-protection guardrails required when using LLMs. The core issue is that LLMs sit outside the cloud tenant in most enterprise AI deployments, which means that data leaves the company's perimeter with every prompt, RAG retrieval, and tool call. Contractual agreements can restrict the data that LLM vendors are allowed to use for training and audits, but they don't stop prompt injection or unintended exposure as company data is often shared to LLMs via natural language queries, APIs, tool and function calls, and MCPs. Sathiesh discusses ways to employ security measures and data filtering at each layer to conform to data security policies and protect the data. 

Topics covered in this episode: Python 3.12.14, 3.11.16, 3.10.21 - security releases Codeberg's AI-code ban tests its role as a GitHub alternative Brett Cannon: what's missing for reproducible builds on PyPI nothing records the source code a distribution came from. direct_url.json captures it when you install from a repo or archive, so the fix is putting the same info in sdist/wheel metadata. recording the build tools. Wheels can already do this via PEP 770 SBOMs in .dist-info/sboms/ - sdists can't, since they're a tarball plus a precalculated PKG-INFO with nowhere to hang extra metadata. Either "don't use sdists" or an sdist v2. Extra extra extra, hear all about it Extras Joke Watch on YouTube Sponsored by Logfire from Pydantic pythonbytes.fm/logfire This episode is brought to you by Pydantic Logfire. It's observability for AI apps from the team behind Pydantic - agents, LLMs, APIs, database, and infrastructure in a single trace, queried with Postgres-compatible SQL. Your coding agent can query it too, through their MCP server. I'll tell you more later. Connect with the hosts Michael: Mastodon / BlueSky / X / LinkedIn Calvin: Mastodon / BlueSky / X / LinkedIn Show: Mastodon / BlueSky / X Join us on YouTube at pythonbytes.fm/live to be part of the audience. Usually Tuesday at 7am PT. Older video versions available there too. Finally, if you want an artisanal, hand-crafted digest of every week of the show notes in email form? Add your name and email to our friends of the show list, we'll never share it. Calvin #1: Python 3.12.14, 3.11.16, 3.10.21 - security releases https://blog.python.org/2026/08/python-31214-31116-31021/ Source-only security releases for the three branches now in security-fix-only mode; release team blamed the European solar eclipse for the timing. tarfile hardening. Multiple path-traversal bypasses of the data filter closed, including a symlink escape that bypassed the CVE-2025-4330 fix; extract() now applies the filter to link targets too. Four fresh CVEs: CVE-2026-2297 (SourcelessFileLoader not using io.open_code() for .pyc), CVE-2026-4224 (expat crash on deeply nested content models), CVE-2026-3644 (control chars in http.cookies.Morsel), plus the completed CVE-2021-4189 fix in ftplib.ftpcp. Quadratic-complexity DoS cleanup across the stdlib: HTMLParser, configparser regexes, unicodedata.normalize(), csv.Sniffer.sniff(), and ElementTree XPath index predicates. Header/injection fixes: CR/LF rejected in HTTPConnection.set_tunnel(), control chars blocked in wsgiref.handlers status, and webbrowser now rejects leading dashes (plus a %action prefix bypass). http.client now caps chunked trailer lines and 1xx interim responses at 100 each - a hostile server could previously hang the client forever despite a socket timeout. Memory-safety odds and ends: stale pointers in lzma/bz2/zlib decompressors after MemoryError, a bz2 stack overflow on reuse-after-error, and bundled libexpat bumped to 2.8.3. If you're still on 3.10, 3.11, or 3.12 - and you extract tarballs from anywhere you don't fully control - this one's not optional. Michael #2: Codeberg's AI-code ban tests its role as a GitHub alternative Armin's article “Codeberg Divides” Armin Ronacher argues that Codeberg's new terms, which prohibit projects mostly written with generative AI, create a vague and difficult-to-enforce boundary. His larger concern is that a democratically governed host can still be unpredictable or ideologically narrow, weakening Codeberg's potential as a broad European alternative to GitHub. The strongest question for Python developers is whether repository hosting should judge legal open source by how code was produced, or focus on behavior and resource abuse. “Mostly generated” is hard to measure in modern codebases where developers mix handwritten code, completions, agents, and generated refactors. Ronacher suggests clearer alternatives: ban all LLM involvement, or target autonomous repository spam, abusive resource use, and low-quality generated contributions directly. Codeberg is free to choose a values-driven community, but that may conflict with being predictable, neutral infrastructure and a serious GitHub competitor. Worth discussing: can open-source communities set meaningful AI boundaries without driving maintainers and projects into opposing camps? Very first search for these terms lands on this page. Codeberg looked like a viable alternative. … Unfortunately, the latest update to its terms of service seems to mark a first step in changing one part I moved there for, namely the “freedom” part. Sponsor: Logfire from Pydantic Your AI agent failed at 2am. Was it the model? A tool call? The database? Most observability tools can't tell you, because they only see part of your stack. Pydantic Logfire sees all of it. One trace across your agents, LLMs, APIs, and database. Down to the infrastructure: services, Kubernetes, and hosts. It's built on OpenTelemetry, with SDKs for Python, TypeScript, and Rust, and it works with any OTel-compatible language. Every prompt, token count, and cost, right next to your vector searches and API calls. You query everything with Postgres-compatible SQL. And so can your coding agent, through the Logfire MCP server. Stop guessing. Read the trace. Pydantic Logfire. AI, it's still just engineering. Visit pythonbytes.fm/logfire today and sign up today. Get 10M records free every month, no card required. You can even click “Onboard with your coding agent” to copy a prompt to have claude or codex integrate Logfire into your app. Thanks to Pydantic for supporting the show. Calvin #3: Brett Cannon: what's missing for reproducible builds on PyPI Framing came out of his 2026 Python Packaging Council nomination - the secure-supply-chain gap he found is that Python has no defined way to do reproducible builds at all. Design goal is zero friction: producers uploading to PyPI shouldn't have to do anything. The work lands on build backends and installers. Gap #1: nothing records the source code a distribution came from. direct_url.json captures it when you install from a repo or archive, so the fix is putting the same info in sdist/wheel metadata. Gap #2: recording the build tools. Wheels can already do this via PEP 770 SBOMs in .dist-info/sboms/ - sdists can't, since they're a tarball plus a precalculated PKG-INFO with nowhere to hang extra metadata. Either "don't use sdists" or an sdist v2. The replay mechanism already exists: [build-system] in pyproject.toml is a defined entry point, so if backends recorded their own environment, you could reinstall and re-run the build. Payoff idea: trusted third parties report successful reproductions back to PyPI, which displays "independently reproduced by X" - surfaced in the index API so installers could prefer reproduced files. Explicitly framed as a perk, not a requirement - roughly SLSA build level 1, no shaming projects that don't opt in. Verbal kicker option: "And don't think pure-Python wheels are off the hook. Something built that wheel, and if that something was compromised, so is your wheel. SolarWinds was a build-process attack." Michael #4: Extra extra extra, hear all about it Python 3.14.7 Upgraded the MCP servers to 2026-07-28 v2 protocols (talk python, python bytes) Got agentsview running synced via postgres Talk Python courses, teams trial offering Talk Python courses, government procurement offering Lean TDD audio book is out Extras Calvin: uv now prefers post-quantum key exchange - https://github.com/astral-sh/uv/releases/tag/0.12.4 Joke: Beware of dog

Bloomberg reported that Anthropic surpassed a $65 billion annualized revenue run rate ahead of a planned IPO. Anthropic, founded by CEO Dario Amodei and President Daniela Amodei, sells access to Claude models via APIs, enterprise products, and cloud marketplaces. The company partners with Amazon and Google, with Amazon committing up to $4 billion and Google providing about $2 billion in financing reported in late 2023. The run rate metric can include usage-based contracts, commitments, and prepayments, and does not disclose margins or revenue concentration. Cloud providers such as AWS and Google Cloud are positioned to benefit through compute consumption and bundled services. Enterprise buyers are shifting to negotiated volume tiers, reserved capacity, and cost controls while comparing embedded AI features to direct API use. If Anthropic files an S-1, disclosures on revenue mix, cloud commitments, and gross margins will shape investor expectations.Learn more on this news by visiting us at: https://greyjournal.net/news/ Hosted on Acast. See acast.com/privacy for more information.

In this episode, Paul Galloway and Craig Jeffery discuss treasury technology debt, the hidden burden created when legacy systems and infrastructure continue to work but increasingly limit flexibility, security, and future capabilities. They explain how treasury teams can assess technology and identify where outdated tools create the most drag, and how to prioritize upgrades without trying to replace everything at once. The discussion also covers ISO 20022, APIs, AI, cloud-native architecture, interoperability, and why treasury should regularly review its technology stack to maintain a practical runway for future change. 2026 Treasury Technology Analyst Report The Hidden Costs of Complexity in Treasury Operations (2026) TMS and TRMS: Choosing the Right Treasury System in 2026 (2026) Treasury Assessments: What to Review and How Often (2026) Timestamps: 00:00 Introduction 01:16 What is treasury technology debt? 02:45 Common forms of technology debt 04:28 How outdated technology impacts treasury 06:44 Identifying and prioritizing technology debt 10:25 Payment standards and ISO 20022 11:31 Reducing technology debt strategically 14:15 APIs, interoperability, and modernization 15:40 Building a sustainable technology roadmap 16:31 Final thoughts 16:48 Outro ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ ABOUT STRATEGIC TREASURER ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ Strategic Treasurer is recognized as a top tier consulting firm in the area of treasury and risk management. Corporate clients, banks, and technology vendors all rely on their industry leading advisory services that are backed by a deep awareness of current needs, practices, and budgeting priorities of treasury professionals through Strategic Treasurer's annual industry surveys and decades of treasury experience. Strategic Treasurer utilizes a senior consultant model where every project is managed by senior consultants with actual practitioner experience in corporate and/or banking roles. Download The Strategic Treasurer: A Partnership for Corporate Growth by Craig A. Jeffery in Kindle, or hardcover. As an Amazon Associate, we earn from qualifying purchases. Visit us today at StrategicTreasurer.com. Or join in the discussion at one of our leading LinkedIn groups.

Has AI actually helped you make more placements this year, or has it just been noise? Before you dive in, this is your last call for The Recruiting Agents Workshop with Seb Sharp, August 25 and 26. Two live build-along sessions where you will create an autonomous lead agent, a placement agent, and connect your entire tech stack, plus recordings, templates, and a week of Slack access to Seb after the sessions end. Grab your seat now: https://the-recruiting-agents-workshop.heysummit.com/ In this episode Benjamin Mena sits down with Dan McCarthy, Senior Talent Engineer at Zapier and one of the founding members of the a16z Talent Engineer Fellowship, to break down recruiting's newest role and how you can step into it before the rest of the industry catches on. Dan's path into this seat is anything but typical. He was a jazz musician, a New York bartender, a wine educator, and a CrossFit gym owner in Brooklyn before Shopify hired him into tech recruiting at 40 years old. He had never heard of an org chart. By his own telling, he led the entire engineering recruiting team in hires his very first quarter. Two layoffs later he landed at Zapier, where he was handed a blank canvas: build our talent intelligence function. What he built instead of dashboards is the heart of this conversation. Dan walks through the talent intelligence MCP he assembled in about a month using completely free APIs, including Indeed Hiring Lab, the Department of Labor, O*NET, the Bureau of Labor Statistics, and WARN Act data, so recruiters walk into every intake call armed with real compensation numbers, competitor hiring activity, and talent pool data. Then he gets specific about the agency version: what a solo recruiter or a two person shop can stand up over a single weekend to change their next client pitch. Benjamin and Dan also get into the ego build problem on LinkedIn and why screenshots of tools built yesterday are making everyone feel further behind than they actually are, the difference between building to learn and building to ship, who is reviewing your code and updating your API keys, whether a three person agency could really bill five to ten million dollars with a builder in one of the seats (Dan's honest answer: possible, but probably not in the next six to twelve months), the sales engineer ratio that may be coming to recruiting teams, and why Dan completely changed his mind about AI interview screens. If you have ever felt too far behind to start building, this episode is your permission slip. As Dan puts it, nobody is behind. This is day one. ⚡ The Recruiting Agents Workshop with Seb Sharp (August 25-26): https://the-recruiting-agents-workshop.heysummit.com/

Before AI, Klaviyo had an insight that changed e-commerce marketing forever: don't show merchants how many emails they sent, show them how much money those emails made. That single shift helped Klaviyo capture 80% market share, earn a cult following among Shopify merchants, and IPO in 2023 with $1.4B in revenue. Now Andrew has to do it again. In this session at SaaStr AI Day, Klaviyo's co-founder and CEO breaks down how he's rebuilding a dominant pre-AI company for the agent era, including some of the most practical AI-at-scale lessons we've heard from any founder this year. What's inside: The Dark Factory. Klaviyo's internal agent system inspired by "lights-out manufacturing," where a team of agents decomposes a product prompt into specs, writes software interfaces, runs tests, handles load testing, and raises a flag only when it's genuinely stuck. One weekend, one agent team, one full prototype. The Tom Brady Rule. Andrew's mental model for LLMs: treat them like a great all-around athlete. They'll never be elite without coaching. The "harness" you build around the model, the domain-specific data, feedback loops, and scoring, is what makes the difference between a POC and a product that works for 200,000 customers. Agents skip onboarding entirely. Unlike human users who take months to learn a product, agents land as power users on day one. Andrew's team now asks their own agents, "What's holding you back?" and uses the answers to build their product roadmap. (One example: the agent discovered AMP interactive email on its own and immediately asked Klaviyo to build the missing APIs.) Agents training agents. For customer-facing agents, Klaviyo takes a feed of real support cases, classifies them, then runs an agent loop that trains another agent on the platform, without any human FDE or SE required. It ships at 50-70% resolution out of the box. APIs are the new UI. Whether your software is brand new or a decade old, if your internal and external APIs are great, agents will figure out the rest. The companies that win the agent era won't be the ones with the best interface. They'll be the ones with the best infrastructure. Three years from now, every business will have an agent you can reach through a URL, a phone number, or an email. The question is who builds the infrastructure underneath it.

In this episode, Zachary Smith, CEO and co-founder of Datum and previously the founder of Packet (acquired by Equinix for $335M) and Voxel (acquired for $35M), joins Liam to explain why the internet is about to undergo its biggest transformation since the cloud. As AI agents, vibe coding, and thousands of new applications flood the web, Zac believes the open internet model we've relied on for decades is breaking down. Zac argues that every person, every company, and eventually every AI agent will need its own private network. He explains why the future internet may look more like the Visa network than today's public web, how digital sovereignty and geopolitics are reshaping infrastructure, and why developers are increasingly relying on dozens of cloud services rather than just the hyperscalers. The conversation also dives into Zac's unlikely journey from Juilliard-trained musician to building and exiting two infrastructure companies, the emotional toll of entrepreneurship, and why he keeps coming back to startups despite already having financial freedom. Key Topics Covered Zach's journey from Juilliard and classical music to building infrastructure companies Building Voxel and selling the company for $35M Starting Packet and its $335M acquisition by Equinix Why AI agents are creating a security problem for the internet Why every person and company may eventually need a private network The difference between the public internet and private internet Why the future internet could resemble the Visa network Digital sovereignty, geopolitics, and the splintering of the internet Why developers increasingly rely on dozens of cloud providers How AI is turning millions of people into software developers APIs, MCP, and the next phase of application architecture Why Zach believes AI agents should only talk to approved systems Open source, network effects, and Datum's long-term vision The emotional side of entrepreneurship and why community matters more than money Episode Timestamps 00:00 Introduction and welcome 00:06 Zach's background: from Juilliard and classical bass to startups 02:44 Building Voxel and the early cloud era 08:44 Starting Packet, raising capital, and the Equinix acquisition 15:28 Why taking time off helped him dream again 18:02 What Datum does and the idea of a network cloud 19:38 Three forces changing the internet 20:41 Hyperscalers explained: Amazon, Google, and Microsoft 24:52 Why new cloud providers are emerging 27:16 Digital sovereignty and the fragmentation of the internet 32:03 Public internet vs. private internet 32:54 Inside the physical "meet me rooms" that connect the internet 39:49 How internet routing actually works 45:56 Why developers use so many cloud providers 48:10 APIs, MCP, and AI agents 51:07 Why the future internet may resemble the Visa network 54:23 Who Datum's customers are, and why Datum is open source 1:03:07 AI agents and the next generation of software 1:07:50 Why Zach keeps building companies, and why he does what he does Connect with Zac: LinkedIn: https://www.linkedin.com/in/zsmith/ Website: https://www.datum.net/ Partner Links Upgrade your AI toolkit: https://www.theaireport.ai/ai-executive-pass Subscribe to our free newsletter: https://newsletter.theaireport.ai/subscribe Join the community: https://community.theaireport.ai/checkout/the-ai-report-welcome-gift?coupon_code=WRTH Learn more about your ad choices. Visit megaphone.fm/adchoices

Topics covered in this episode: Claude Code /insights Post-quantum crypto lands in Python MCP goes stateless — and FastMCP gets renamed inshellisense - IDE style command line auto complete Extras Joke Watch on YouTube About the show Sponsored by Xweather Xweather combines enterprise-grade weather intelligence with agent-ready APIs, natural language capabilities, and an MCP server so your agents can adapt workflows, automate responses, and make better decisions based on real-world conditions. Michael will tell you more about them later in the show. Get started for free at pythonbytes.fm/xweather Connect with the hosts Michael: Mastodon / BlueSky / X / LinkedIn Calvin: Mastodon / BlueSky / X / LinkedIn Show: Mastodon / BlueSky / X Join us on YouTube at pythonbytes.fm/live to be part of the audience. Usually Tuesday at 7am PT. Older video versions available there too. Finally, if you want an artisanal digest of every week of the show notes in email form? Add your name and email to our friends of the show list, we'll never share it. Michael #1: Claude Code /insights Michael's Insights: michael-kennedy-claude-code-insights-2026-08-09.html Be careful sharing these outputs, they include details references to your projects, errors, security findings, etc. ;) /insights reads your last 30 days of local session transcripts and hands back an interactive HTML report on how you actually work. One command, zero setup: type /insights in a session, or run claude -p "/insights" from the shell for a non-interactive version that just prints the path Reads what's already on disk: pulls session logs from ~/.claude/projects/, skipping agent sub-sessions and anything under 2 messages or 1 minute Project areas: clusters your sessions into themes like "CLI Tooling" or "Documentation" with session counts Friction analysis: categorizes where things went wrong by root cause - and quotes your own prompts back at you Interaction style: tells you whether you're a delegator or a micromanager, plus which workflows are worth doubling down on Actually actionable: suggests concrete CLAUDE.md additions and Claude Code features you're not using The catch: Haiku does the per-session classification, so the first run takes several minutes; results cache to ~/.claude/usage-data/facets/ and the report lands at ~/.claude/usage-data/report.html Calvin #2: Post-quantum crypto lands in Python pyca/cryptography 48 ships ML-KEM (key establishment) and ML-DSA (signatures) — NIST's post-quantum standards, now one pip install away. Big deal because it's the 11th most-downloaded package on PyPI (~1.2B downloads/month) and sits under Ansible, Certbot, Airflow, and paramiko. No PQ there, no PQ anywhere in Python. Trail of Bits did the work (Rust bindings, cross-backend API, tests, AWS-LC backend support), funded by the Sovereign Tech Agency. Timing tracks a June 22 White House order setting federal deadlines: PQ key establishment by end of 2030, PQ signatures by end of 2031. Not a drop-in swap — the wire sizes explode. ML-DSA-65 signatures are 3,309 bytes vs Ed25519's 64; ML-KEM-768 public keys are 1,184 bytes vs X25519's 32. Hardcoded field sizes and length prefixes will bite. API looks like the existing asymmetric primitives, except ML-KEM is encapsulate/decapsulate rather than a Diffie-Hellman exchange. SLH-DSA (the hash-based conservative backstop) is still in progress. The primitives are here, but protocols haven't caught up — so you won't be running post-quantum Certbot this week. Sponsor: Xweather You're using agents that can write code, summarize documents, and automate workflows. But they're missing one thing: awareness of the world around them. This is where today's sponsor, Xweather comes in. Xweather combines enterprise-grade weather intelligence with agent-ready APIs, natural language capabilities, and an MCP server built for tools like Claude, Codex, Copilot, and modern IDEs – so your agents can adapt workflows, automate responses, and make better decisions based on real-world conditions. Backed by Vaisala, whose instruments fly on NASA missions to Mars, Xweather delivers trusted data and unique insights that go beyond conditions to actual impact – from real-time lightning strikes to road surface forecasts. Start with 15,000 free API calls each month and pay only for what you use as you grow. Xweather is your full weather stack, for developers by developers. Start building for free today at pythonbytes.fm/xweather. The link is in your podcast player's show notes and on the episode page. Thanks so much to Xweather for supporting Python Bytes. Calvin #3: MCP goes stateless — and FastMCP gets renamed From Philipp Acsany over at Real Python The 2026-07-28 spec landed July 28 and the Python SDK shipped 2.0.0 the same day. Biggest rewrite since MCP launched, and it's breaking on purpose. Context for scale: the Tier 1 SDKs are pulling close to half a billion downloads a month, with TypeScript and Python each past a billion total. The headline is the stateless core. The initialize/initialized handshake and the Mcp-Session-Id header are both retired — protocol version, client identity, and capabilities now ride in _meta on every request, with an optional server/discover RPC if a client wants capabilities up front. Any request can land on any instance behind plain round-robin, no shared storage. Server-initiated calls are the hard part of the migration. Sampling, elicitation, and roots/list no longer call back to the client; instead the server returns resultType: "input_required" and the client retries with inputResponses attached. Multi Round-Trip Requests, MRTR. Also: Mcp-Method and Mcp-Name are now required headers so gateways route on headers instead of cracking JSON bodies, and missing-resource errors move to standard 32602. Deprecation sweep with an actual policy behind it — Roots, Sampling, Logging, and the legacy HTTP+SSE transport all deprecated with a twelve-month minimum offramp. Tasks graduated out of the experimental core into a real extension, which is what the formalized extensions framework was for. MCP Apps is now an official extension too, so a tool call can return sandboxed interactive HTML. Auth picked up RFC 9207 issuer validation, issuer-bound credentials, and a shift from DCR toward CIMD. Python SDK 2.0 is where it gets personal: FastMCP is now MCPServer, no alias, no shim. McpError → MCPError. Wire types went snake_case (is_error, input_schema) and moved to a standalone mcp_types package, with mcp.types kept as a permanent alias. One Client object replaces the old transport + ClientSession + initialize() stack. httpx became httpx2. Sync handlers run on worker threads now, so asyncio.get_running_loop() raises inside them. The good news: one MCPServer serves both protocol eras, so 2025-era clients keep working with nothing to configure, and a Resolve(fn) parameter lets one tool body cover MRTR and the old path. 1.x is maintenance-and-security-fixes only — pin mcp>=1.28,

Waiting three to five days for funds isn't just annoying, it quietly reshapes entire business models. I sit down with Arpit Goel, CEO of Root, to unpack why money movement in the United States still depends on slow, multi-hop workflows and how that slowness creates “float” across payroll, staffing, marketplaces, insurance, and the creator economy. When settlement is delayed, someone benefits, and Arpit makes a sharp case that incentives are the real blocker, not the lack of technology.We dig into what Root is building in the instant payments space: a non-custodial, bank-to-bank infrastructure layer that aims to deliver true real-time payments where funds settle in seconds and can be used immediately. Arpit explains the “Zelle for businesses” idea, why it matters for enterprise disbursements, and how programmatic treasury APIs can help companies ship faster without spending a year integrating directly with a bank. We also talk about the current US real-time rails, including FedNow and RTP, and why adoption is accelerating as standards and competition push the ecosystem forward.Then we zoom out to the trends payments leaders can't ignore: agentic payments, fraud and liability, strong customer authentication, and the stablecoin-to-bank convergence as regulation catches up. Arpit's final message is direct: compliance is a follower, and security needs to be years ahead as AI-driven attacks scale faster than humans ever could. Subscribe for more conversations like this, share the episode with a payments leader, and leave a review so more builders can find it.

According to Gartner, worldwide AI spending is forecasted to increase 44% by the end of 2026. Companies are investing in AI, and they are investing heavily. But knowing where and how to invest isn’t easy, especially with what feels like a million different AI tools out there and a million more different ways to build your own. So how do you figure out what to build, what to buy, and which investments will help you move the needle for your business? Riley Rogers: Hi, and welcome to the Win/Win Podcast. I’m your host, Riley Rogers. Join us as we dive into changing trends in the workplace and how to navigate them successfully. Here to discuss this topic is Cody Sims, head of commercial brand at Cox Communications. Cody, thank you so much for joining us today. Super excited to hear your thoughts on this one. Before we dive into what is quite a loaded topic, could you tell us a little bit about yourself, your background, and your role? Cody Sims: Yeah. So, hi, I’m Cody. I’m the head of commercial brand for Cox Communications, and it’s kind of crazy how I came into this role. So I actually started my career when I was 15 and was installing phone systems for my dad’s phone company. And after that, I had actually had two parts of what I thought was what I wanted to go into, and that was either musical theater or physics, because those were two things I really had a passion about. And when I got into college and had musical theater as my major and physics as my fallback, I realized that both of them left a part out of what I really enjoy. And so I ended up actually landing in marketing because it’s both analytical and creative, and that has served me really well over the years. So, across Cox, I have done all kinds of things from product management to market development to pricing to competitive analysis, and now in the brand world. It’s given me kind of a 360 view of the entire business from a marketing lens. I would say that I’m pretty much a transformation leader. I really enjoy breaking things and building them up new again. So, AI is happening right at the right time for me. RR: I love that story, and I love that it’s taking you to a place that especially now is getting more and more technical, more and more analytical. I’m very excited to get into that transformation leader side of things. But before we do, can you paint a little bit of a picture of your sales environment? CS: Yeah. So when I first came to Cox, it was very similar to most of what you would call a CLEC, or a competitive local exchange carrier, which is primarily internet service, voice services, and obviously because it was Cox, some cable TV services that were the triple threat. That was kind of what in the early 2000s was kind of the way that they went to market. But over time, the team at Cox realized that in order to stay competitive, they had to add to the portfolio to make sure that they were providing value to their customers, and I’m sure many would understand that and have gone through similar transformations. And so we had acquired several different other companies that added to our portfolio, and developed some of our own products, and over time that turned into a lot of products. But it’s not just 70 products. It’s 70 products, it’s nine customer segments that we have from a segmentation perspective. It’s six distinct buyer personas, industry verticals, what’s serviceable at that address. So you take all of these different components and it’s almost like three-dimensional chess for the salesperson. The way that I like to think about it is that the sellers, what they really need and what their challenge is, is that they’re not looking for specs. They’re looking for what are the business outcomes that my customer is trying to achieve, and then what do I have from my portfolio that will help them to achieve those results? So it’s no longer a world where they can memorize everything and know every product in and out, and be the technical expert. They really do have to have tools and systems that help them to have the right knowledge at the right moment for the right person in the right place. RR: Yeah, there comes a point when the human brain just can’t contain the context and the expertise that you need. So when you can’t ask for expertise, what you can do is provide, to your point, that just-in-time support. And one of the things that you alluded to was that that’s where you kind of started some of that AI investment as a way to bridge that gap. You’ve given us a little bit of a taste, but what kind of motivated that early initiative? CS: Well, I would say that, over time what we discovered was that we couldn’t keep track of all of our marketing materials, collateral, all of the pieces of information in just files, formats, and putting it online into a here’s-an-accessible-library. Because the library just becomes bigger and larger and more difficult to manage. But I would say that we didn’t set out to do AI. We didn’t sit down and say, “Oh, hey, AI looks cool. Let’s make sure we’re doing it.” We needed to transform our go-to-market strategy so that we were more nimble, we were more competitive, and that we could deliver the kind of experience that our customers were asking for. And so AI was the mechanism that would help us to get there. But what really triggered this whole thing was what I mentioned earlier, was our segmentation. When we sat down and said, “Let’s rebuild the way that we look at our audience segments,” and we did that based off of what is the value to Cox of each of these customer profiles, and then what is the technology sophistication of that client, of that business. And that intersection allowed us to create our nine different segments that we were working on. And so when we did the math, when we looked at all of that information and all of the things that we needed to be able to provide to those segments, we realized this was quickly going to turn into something that was far beyond any marketer’s ability to do. But what we knew is that the Gartner information we were tracking said that personalization was having much higher returns on the way that people respond to information. And not only that, but if you do personalization and you get it wrong, if I call you and I say, instead of, “Hey, Riley,” and I say, “Hey, Jonah,” you’re like, “Hmm, nice try.” So personalization is really important to being successful, but getting it right is even more important. So we realized that we needed to have some radical partnership between our marketing, AI, and sales teams, that we needed to make sure that this was not just an IT project, that we were going to go and pull a bunch of requirements together and everybody would be like, “Oh, hey, here’s this new tool. Everybody figure out how to use it.” And it wasn’t necessarily about optimization. It was about transformation, the way that we go to market, the way we think about our customers, and the way we show up. So I would say that AI definitely was part of the solution set, but we had to look ourselves in the mirror and say, “It’s time for us to actually think about this in a completely different way.” RR: That distinction comes through very well, and I think is very important because oftentimes when you’re in kind of the scramble to be keeping up with the market, keeping up with your competitors, there is this urge to just tack on AI because we have to. But when it’s not strategic and it’s not built into the things that you’re actually doing, to your point, it’s we put together some specs, good luck using it. But instead, now it’s something that’s really built into the way that you work. I would love to hear a little bit more about that specific use case, especially given the fact that a lot of teams are running into that question of how do we use AI and can we just build what we need ourselves? Given that you’ve done the math, answered the question, I’d love to hear how it worked and kind of where you landed. CS: It’s very easy to fall into the trap of, “Hey, everybody, here’s AI. We put it on your computers, now go use it.” And so then everybody starts using AI to try to figure out, how does this help me in the job that I already do, in the role that I already do, in the processes that I already do? And so then it really limits the impact that it can have on the business and the performance because either, A, you have a handful of people who are really smart, and they go crazy with it, and they create their own thing, or you have a bunch of people who are looking at it going, “Okay, came up with some ideas, but I still have to do my work.” What you end up with is there’s no standard. There’s no flag running up the hill to say, “Everybody follow me. Let’s go do it this way.” So, it required both the yes, we had to make sure that the teams were bought into using AI, but we also had to have a standardized way of approaching how we deploy AI. And that brought us to the question of do we buy or do we build? And because there were so many different parts of what kinds of functionality we needed, it wasn’t the same answer for every one of those needs. So in some cases, we have a tool, we have a partner, they already have AI integrated into their platform, let’s go see how we can use that. In other cases, and I’ll give you an example, in the case of content generation, that is where we started with our AI journey about a year ago. We sat down and started interviewing and reviewing all of the different providers who can do content generation. Every one of them had a different approach to content development, content generation, which were all very good, and they attempt to make sure that they are covering as much of the marketplace as possible. And so sometimes when you buy that, you end up with features maybe that you don’t need, and you also have to still go through the process of integrating those platforms into your security posture. So us being a connectivity provider for governments, for major corporations, enterprise carrier grade, we have a very, very strict and strong security policy, which means that when we bring new vendors on, it takes a lot of time and a lot of effort and a lot of back and forth. And so what we found in certain cases, it was actually better for us and more beneficial for us to build the actual platforms that we needed for that particular use case. But like I said before, in other situations, we found that there was a partner who we had who already had AI integrated into their platform, and so they were already part of our security posture. They were already inside of our ecosystem. So the question of build versus buy really had to do with time, had to do with return, and it had to do with the security measures that we had to put in place. RR: Thinking about in addition to those factors, when you’re evaluating these things that you outlined, time, potential cost, security, how are you kind of doing that ROI math to say one is going to be better than the other? CS: There’s several different parts of that. And like I mentioned, we wanted to make sure that we were following our AI strategy foundation that said, we don’t want to introduce more and more vulnerable access points. And so it’s important for us to make sure that we are all coming together with everyone across the Cox leadership team according to who are the vendors that we feel the safest with, that we can go set up and make sure that we are pulling together the best of the breeds. The assessment, like I mentioned before, is what is the value that we’re returning to the business in terms of revenue generation, new customers, cost savings in terms of not necessarily just reducing people’s time, but redeploying people to doing other important tasks. And then what are the things that we are doing that help us to keep the system all working together? So, revenue generation, cost deferment, and then keeping a cohesive connection between all of the different platforms. So some of the things that we looked at from our comparing vendors versus doing DIY, is there a maintenance tail that goes in this? So if we build it, what does that look like in 18 months? How much more people do we have to have to support it? Governance and observability, do we have the permissions, the versioning, the audit trail, all of the parts for discovering what is needed and then able to see it and observe it as we go? Interoperability, as I mentioned before, really important between different platforms that we have, that those APIs and MCPs all work together. And then whose roadmap is this? Is this our roadmap? Is this the IT roadmap? Is this the vendor’s roadmap? If we know where we need to go, is there anything that’s getting in our way of being able to get there? And then of course, obviously the speed to value against the cost of being wrong. RR: And so hearing you outline this very comprehensive list of considerations, you can start to understand why it starts to feel complicated and really hard to tackle. To your point, it’s been a year of figuring it out since you started developing that very first use case. I’d like to go into a little bit of detail about the evaluation piece and deciding what vendors you felt safe with, that you were excited to partner with and continue to either use or build upon as you’re developing your AI strategy in alignment with your business transformation. One of those that you landed on was using Highspot’s MCP server to support some of the workflows you wanted to spin up. How did you make that decision and why did that feel like the way to go? CS: Well, as I had mentioned before, as we had gone through our history of, here’s a library of a whole bunch of stuff and everybody’s trying to find the right item, and it just was such a headache to make sure that we were always getting the right information to the right customers at the right time. And not only that, but we had no real clear feedback about how it was performing. And so at that time, which I believe was in the 2015 to 2017 timeframe, is when we had first started our relationship with Highspot to help us better catalog the library, make it more searchable and usable and referenceable for the sellers to be able to share information and track the information, make sure that it was the most relevant and recent, and then help us to understand what’s working and not working. So all of that was already in place before we even started the AI conversation. And so as we were doing our work around our go-to-market roadmap, we started with content because it was probably the easiest place for us to use AI to generate content, and that looked like a two-layered approach. We had what we called a knowledge base, which is formally putting into AI rules that can be read by AI around all of our standards for brand, for legal, for segment definition, for product information, for pricing and promotion information, industries, verticals. All of that was put at this knowledge base foundation layer. And then we built the content generation engine on top of that, where each of the agents within that tool would go find what it is that the marketer was asking to do, compare it against all the information in the knowledge base, the brand standards, all of those good things, and then produce the content piece that the marketer was asking for using that foundation layer. However, once we got that moving and going, we realized that that level of personalization for marketing could be even more valuable and even more specific when used by a seller. But in order for that to work properly, the seller had to have access to a large range of information all at the same time, including any of the buying signals or online signals that we had through some of our lead generation partners, any of our information that we have within our own systems, like when was the last time they called into billing or when was the last time that they had an outage or what is their general sentiment that the customer has right now. And then all of the information about their current services, their current products, all the things that are going on in their world. But then once we have all of that information, we have propensity to buy, propensity to churn, propensity all these modeling, now we need to be able to talk to them and provide a recommendation to the seller that says, “Here’s what we recommend you use, what you should say, how you should set it up.” And all of that was inside of Highspot. And so we realized again, we could look at this and say, “Are we going to go buy a new platform? Are we going to use a platform we already have or are we going to go build something new?” And obviously when we looked at the Highspot platform, the MCP servers, and the way that it was laid out and set up already, we knew that that was the right path to go. So what we had started with was the content engine, then we went into a sales enablement engine, and as part of that sales enablement engine, the only way for it to work properly was for us to bring in the Highspot MCP service. RR: And how has that been working so far for your sellers? As you’ve rolled this out, how has it been used? Any anecdotal feedback you’ve heard? CS: It’s pretty funny because we have done multiple rollouts of sales enablement platforms over the years, and as anyone who’s ever tried to roll out new sales items and new sales tools will say, it takes time, it takes consistency, messaging over and over. But in this particular case, when we went out and did our roadshow with all of the sellers and sat down and showed them how the new tool worked, there were so many positive responses, and the adoption was much faster than most of our previous releases of other types of products. And I think that the reason why is because it was bringing together all of those pieces of information that I mentioned before and bringing in the Highspot information that they were already very familiar with. And in our world, we call it the sales asset manager, SAM. And so they were very familiar with SAM and then this new tool with the AI capabilities built into it. Now it’s specifically just telling them, “Here’s what you should do. Here’s the way to lay it out, and here’s all the content to talk to the customer about in what order.” And it took a lot of the burden off of them to research, go find a piece, start to build a story in their head, try to build a deck, and then think about what are they going to share with them in what order. So it’s been a huge benefit to the sellers. They’ve loved it. RR: Yeah, that’s such a strong signal when adoption doesn’t feel like a push and more of a grab. Curious if there are any other AI or agentic connectors that you’re pairing with Highspot in another AI application that you think would be interesting to share? CS: We have basically six different programs or parts of our roadmap, and we’re calling them AI modules, and then they work together in different components for different functions that need to be done. So as I mentioned, we have the knowledge base that is the base. Then we have the content creation tool, which we call CAMI. So it’s Content Automation Marketing Intelligence, and that has everything that is needed to produce and create new pieces of content, and then those content pieces are either generated in emails or things like that. A lot of them actually are put into the Highspot tool. And then we have what we call SAMI, which is the Sales Automation Marketing Intelligence, and that is the tool that integrates directly with Highspot to make the recommendations to the seller based off of all of the other information, the 360 view of the customer. We also have what’s called Livia, which is the Lead Validation and Enrichment. The tool uses all of these multiple different access points and different vendors to pull information about that particular contact to validate that it’s accurate, so that by the time it gets to the seller and they’re going to go do a pitch, they have a lot more confidence that who they’re talking to, the business, and it’s at the right address, and prevents them from wasting time. And then, of course, Highspot is such a critical part of how that story all comes together because it’s capturing all the content that’s being created by CAMI, and then the AI that comes from Highspot is infusing into the SAMI tool that the sellers are using. It’s an interesting thing because somebody might say, “Well, you’re not really using Highspot, you’re using SAMI.” And the reality is, well, yes, I am using Highspot because Highspot is feeding all of that into the SAMI tool. There’s a whole bunch of other stuff we add into that for flavoring, all of the information about the customer so that the seller has a 360 view, but that just sets it up. The what do you do next is what’s coming out of Highspot. The next phase of this that we’re going to is a fully agentic approach to our marketing and sales engine. And what that means is that today, most of the work that’s being done is a marketer who is saying, “Here’s what I need to go get done. I’m going to use AI to help me go do it.” We’re going to flip that script, and we’re going to say, the agents that we create are going to do the work, and the marketers are going to instruct the agents on how to do that work properly and watch it and govern it. That will then accelerate for the sellers as well. RR: We’ve heard a little bit about what’s been built in the last year, but it’s, again, to your point, crazy that that’s one year of building, thinking, strategizing, and it’s come to this point. When you look across all of that, what has changed for your sellers and for the business? CS: Well, I would say the first thing is, is that sellers are now able to focus on what they’re really good at. What I mean by that is their confidence is shifted to focus on outcomes and value. They are now able to build trust and provide value, which is honestly what all of our customers, especially our business owners and decision makers are looking for. And then for the marketers, it’s no longer about building a queue, trying to figure out what is the message that’s going to hit the most people with the most response. This idea of efficiency for media or efficiency for marketing materials. It’s like, what is the one message I can send to a million people and have the most response? Well, now you actually can flip that on its ear and say, “I’m going to personalize it at scale.” So that is super exciting. And then the last thing that I would say is that consistency became structural. The same knowledge base, the same rules across every surface, making sure that our content is clean, correct, built on the same policies and rules, but is personalized. Doing those two things at the same time is very tricky, and being able to do it with AI is the only way we could get there. RR: Curious if you’ve seen any sort of measurable returns. CS: Our lead accuracy, like I mentioned before, moving from that 13 to 18% all the way up to the 95th percentile. We have campaign speed to market of improvement of 55%, meaning the amount of time that it takes us to get to market is cut in half. The marketing content teams are 40% more productive, which means they’ve been able to redeploy their time for 40% of the time that they spend at work on other projects, which is amazing. Our conversion rates are up, our driving net new revenue is up, and we have seen material improvement in click-through rates and conversion rates when we are more specific and personalized to the audience. So Gartner was right. Yay. So that’s been really good. And I would say that part of the reason why I think that, at least for part of what we did, doing it as a build ourselves was wise, is because we learned so much by going through the process of just banging our shins on the corners and running into cabinet doors that were open, and we’re just like, “Oh, wow, that was, I did not see that.” So it’s been a huge learning process, a very, very intense learning process, but we’ve all had a really good sense of humor and amusement and just, we are having a ton of fun. RR: And I think that’s one of the more encouraging things to hear. Is that nobody starts perfect, and you just have to build your way up to good. And once you get there, you start to see again, like those measurable improvements. But it is a process. So I guess the message there is stick with it. Which I think kind of feeds into that last question I have for you, which is for anybody who is running into this question, hitting their shins on all of these problems, how would you recommend they approach the question of building, buying, blending some things together when they’re thinking about their AI investments? CS: Well, I would say the first thing is you have to look in the mirror and be real with yourself and say, “Is my processes and workflows working? If I blew up my entire go-to-market, I blew up all my processes, what would it look like?” And don’t start with a tool. Start from a place of what would serve me best. The other part of it that I would say that Highspot did really well is because of the MCP product, I was able to look at it as how am I using this from a plumbing perspective, not just a judgment perspective. And what that means is that it worked well with the strategy and the AI strict rules that we had built for ourselves. Highspot, kudos to Highspot, built a platform that is trusted and that works well with all of the other components that we had flying around, whether it was Salesforce or AWS or even our Accenture development team being able to use the components and pieces to connect to the whole ecosystem. Then the other thing I would say is that even though we’ve been doing this for a year, a year is like eons in AI’s time. It was every other week there was something that changed, something new, something shifted. So you have to go into it with this idea of you have to prepare yourself that this is how I set it up now, but I might have to change it tomorrow, and just be okay with that. So my answer for build or buy, my answer is both. Build the things that make sense for you and where you have the resources and when it’s the right fit. But definitely buy when you are in a partnership or when you have someone that you already know that you can trust. RR: Very pragmatic. That’s kind of the only way to do it. One thing I’ll say, I know I am walking away inspired, and I can imagine our audience is going to as well. So Cody, thank you for the time. I really, really appreciate it. It’s been so wonderful to hear a little bit more about what you’re building. CS: No, I love it. And the reason why this is great for me is that it forces me to think back on this journey that we’ve been on for the last year and really consider what is it that has brought us to where we are, what are the things we’ve learned, and then, maybe how are my bruises doing? RR: Well, thank you for the time again. And to our audience, thank you for listening to this episode of the Win/Win Podcast. Be sure to tune in next time for more insights on how you can maximize go-to-market success with Highspot.

Podcast: Exploited: The Cyber Truth Episode: When Cybersecurity Becomes a Safety Issue: Protecting Autonomous and Critical SystemsPub date: 2026-08-06Get Podcast Transcript →powered by Listen411 - fast audio-to-text and summarizationIn this episode of Exploited: The Cyber Truth, host Paul Ducklin is joined by RunSafe Security Founder and CEO Joe Saunders and returning guest Hemanth Tadepalli, Senior Cybersecurity and Compliance Manager at May Mobility, to explore why cybersecurity and safety are becoming inseparable as software takes control of vehicles and other critical systems. Drawing on his experience securing autonomous mobility platforms, Hemanth explains how connected vehicles have expanded the attack surface across software, cloud infrastructure, APIs, sensors, AI models, and supply chains. Joe explores how organizations can build resilience into software, protect legacy and resource-constrained systems, and move beyond the costly cycle of finding, patching, and deploying fixes. Together, Joe and Hemanth discuss: Why cyber-physical systems require a different security mindsetHow autonomous vehicles connect cybersecurity with physical safetyWhy compliance should be a baseline—not the end goalHow AI is changing both cyberattacks and defenseWhere software resilience and runtime protection fitWhy public trust depends on consistency, transparency, and safetyHow security, engineering, and compliance teams can work together Whether you secure autonomous vehicles, industrial equipment, medical devices, or other critical systems, this episode offers practical lessons for building software that is secure, resilient, and worthy of public trust.The podcast and artwork embedded on this page are from RunSafe Security, which is the property of its owner and not affiliated with or endorsed by Listen Notes, Inc.

Show DescriptionWe tackle a listener question about why junior front-end devs are expected to know so much beyond HTML and CSS, then get into Dave's real-world fix swapping old WebKit line-clamp hacks for the new CSS line-clamp property and the subtle bugs that kind of truncation work can cause. They also debate whether AI tools like Claude or Copilot deserve co-author credit on commits, gripe about agent-only coding workflows that skip linting and hide what's actually happening, and wrap up looking at new platform features like the navigation API and a proposed CSS-based routing approach using @route and @view-transition. Listen on WebsiteWatch on YouTubeSponsorsNotionWrite custom tools for Notion Agents that generate assets, query live data, and hit any API. Listen for incoming webhooks from any app, then run workflows with Notion Agents, pages, databases, and external APIs. All of this, on a hosted runtime. Workers are isolated sandboxes managed by Notion, so the code behind your syncs, tools, and workflows runs on our infra instead of your servers.

If you've ever felt like AI is something that's happening to your organisation rather than something you're steering, this episode is for you. James Drury spent a decade rising through the ranks of one of the Middle East's original e-commerce agencies, eventually running operations and AI strategy end to end. What he brings to this conversation is rare - a genuine understanding of how AI works at the enterprise level, and an equally sharp eye for how solo operators and small teams can use it to punch well above their weight.This is a practical episode. No hype, no hand-waving. Just how to actually build something useful.Three key areas we covered:✳️ The free thinker versus factory worker distinction - not everyone in your organisation will go down rabbit holes with AI, and that's fine. The mistake is expecting them to. Find your free thinkers, let them build the systems, and let your factory workers use those systems to produce.✳️ Context is the unfair advantage - what separates a mediocre AI user from a genuinely effective one isn't the tool, it's the context. James's approach of running a personal interview with Claude - talking for an hour or more about your goals, your business, your working style - and embedding that into your folder structure is one of those things that sounds obvious once you hear it and completely changes how you work.✳️ AI as strategist and operator - the goal isn't to use Claude for one-off tasks. It's to wire it into your actual systems via MCP connections and APIs so it can think alongside you as a strategist and execute as an operator. James does this with his accounting, his content system, his calendar - the lot.James's three amplifiers:✳️ Manage your internal feed - be ruthless about what you consume. The algorithm is not your friend. Curate what's coming in - business, gym, whatever fires you up - and protect your mental state the same way you'd protect your calendar.✳️ Decide on your trade-offs - if you want abnormal results, you have to make abnormal choices. That means being clear about what you're optimising for and being at peace with what that costs. Less scattered, more locked in.✳️ Learn AI, every day - not in a burst, not when you feel like it. One session a day, cap your tokens, build something. Do that for six months and you won't recognise how capable you've become.If this episode sparked something, follow or subscribe so you don't miss what's coming next. And if you've got a guest recommendation, I'd genuinely love to hear it.Timestamps00:00 - Introduction01:55 - James Drury's background: from junior analyst to COO and Chief AI Officer05:24 - Free thinkers vs factory workers: finding your AI heroes in a large organisation09:30 - Why AI amplifies what's already there - systems first, tools second11:14 - How to drive AI adoption through cultural pull, not top-down push14:32 - How personality shapes the way you use AI - and where to start15:50 - Setting up Claude Code: the context interview and building your AI brain19:28 - Daily strategy sessions with AI: how James uses Claude to decide what to execute22:36 - MCPs, APIs, and connecting your tools without the technical terror25:15 - Decision fatigue and letting Claude make the call27:00 - Building skills in the open: the video game analogy for AI learning30:31 - How to reorganise a messy Claude Code folder structure33:40 - Voice vs typing: what works for different kinds of AI users37:00 - AI slop vs AI-supported content - there's a difference38:30 - Building a content system: design documents, image generators, and automation45:35 - Hiring for what you're not good at: the case for a creative director47:13 - One tool recommendation: Motion AI for time blocking and project management50:58 - Image generators: James's pick and why55:16 - Amplifier 1: manage your internal feed58:14 - Amplifier 2: have fewer commitments, get obsessed with the right things01:01:15 - Discipline defined: how far you fall, how fast you climb back01:03:13 - Amplifier 3: learn AI, every single day01:05:40 - Where to find James Drury and Tattoos and Typewriters podcast----Get your copy of my Personal Brand Business BlueprintIt's the FREE roadmap to starting, scaling or just fixing your expert business.www.amplifyme.agency/roadmap----Subscribe to my Youtube!! Follow on Instagram and Twitter @bobgentleJoin the Amplify Insiders Facebook Community : www.amplifyme.agency/insidersPlease take a second to rate this show in Apple Podcasts. ❤ It will mean a lot to me.Mentioned in this episode:Signature StudioGet started today - signtaturestudio.me

CoinDesk reported that Mastercard agreed to acquire stablecoin firm BVNK for about $1.8 billion, signaling a push to embed on-chain settlement into mainstream payments. BVNK provides APIs for merchant acceptance, cross-border payouts, and treasury operations, with compliance features for onboarding and monitoring. Mastercard has built crypto capabilities through its 2021 acquisition of CipherTrace and its 2023 Multi-Token Network pilots. Regulatory frameworks such as the European Union's Markets in Crypto-Assets and ongoing United States policy debates will shape execution. Competitors including Visa, PayPal, and Stripe are advancing stablecoin initiatives. Founders should track network roadmaps, model settlement and fee impacts, and prioritize compliance ready integrations as stablecoin rails scale.Learn more on this news by visiting us at: https://greyjournal.net/news/ Hosted on Acast. See acast.com/privacy for more information.

Our Head of U.S. Public Policy Research Ariana Salvatore explains how U.S.-China tensions, export controls and domestic regulation are reshaping where AI is built, who controls it and what investors should watch.Read more insights from Morgan Stanley.----- Transcript -----Ariana Salvatore: Welcome to Thoughts on the Market. I'm Ariana Salvatore, Head of U.S. Public Policy Research at Morgan Stanley. Today, a look at how government is increasingly determining the future of AI in the U.S. – from where it's built to which technologies US companies and consumers can use. It's Friday, August 7th at 10am in New York. AI is rapidly reshaping the economy and society, so this is a pivotal moment for government to consider the rules governing that development. The first area to watch is technology restrictions, particularly in the context of U.S.-China competition. Now, for much of the past decade, the government's approach has been to restrict a relatively narrow group of technologies with clear national security implications while maintaining broader commercial ties. But as export controls spread across more sectors of the economy and AI moves from software into physical infrastructure, the definition of what qualifies as national security has become broader. The Department of Commerce could, for example, expand the entity list. That would require US cloud providers, software companies, and model marketplaces to remove or stop supporting models tied to designated Chinese developers. Congress could then make those restrictions more durable through things like the annual defense bill or other policy vehicles. We're keeping an eye on several legislative proposals, like the AI Overwatch Act, which would tighten controls and give congressional oversight around exports of the most advanced AI chips; and the MATCH Act, which would extend restrictions further upstream to semiconductor manufacturing equipment and seek closer alignment with allied producers. These measures wouldn't directly ban Americans from using a Chinese model, but they could constrain China's ability to train future frontier systems. But it's not just the US that could impose a set of restrictions. China has a parallel set of tools focused more on integration and market access. Regulators could block four models or APIs. They could require locally controlled deployment. They could impose Chinese data and content standards or use cybersecurity and entity list authorities to promote domestic substitutes. The likely result is an increasingly distinct pair of AI ecosystems. That's our two worlds thesis in practice. Over time, we think that means a bifurcated global AI market into separate technology ecosystems. That looks like the U.S. relying on export controls, allied supply chains, and largely closed frontier model platforms, while China emphasizes domestic hardware, open-weight models, subsidized compute, and localization. Over time, that bifurcation could produce different chips, models, standards, data rules, and distribution channels, while third countries navigate between the competing stacks. The second area to watch is domestic regulation. Today, the landscape is pretty fragmented. States are moving first on certain specific issues, including automated decision-making and child safety. Now, at the same time, Congress is confronting competing objectives from industry, consumer groups, and national security officials. So far, we think the evidence suggests that the administration's preference is for a light-touch approach, a largely voluntary national framework rather than a broad new licensing regime. But it's also moving toward more direct oversight of the most advanced models. That includes the possibility to play a more active role prior to model release to ensure that certain protections like cybersecurity and intellectual property are met. Publicly outlined priorities from industry seem to broadly overlap with that approach: a consistent federal framework, clearer liability standards, access to data, compute, and power, and copyright rules that don't materially limit model training. But of course, the industry isn't monolithic. There are some important nuances between frontier developers and other players. So, what does all this mean for investors? The government's reaction function will be critical to the way AI is developed and diffused throughout our society in two key ways. First, we see regulation altering not only the pace, but also the geography of AI infrastructure. At the same time, we think these constraints could strengthen the investment case for bottleneck solutions like on-site power generation, fuel cells, storage, and more. Second, greater technology bifurcation supports investment in parallel supply chains. The key takeaway here is that the government is no longer simply regulating the industry from the sidelines. It's helping to determine how fast AI develops through domestic rules, where it develops through infrastructure, permitting, and sovereign AI policy, and which technologies are accessible through export controls and market access restrictions. Thanks for listening. If you enjoy the show, please leave us a review wherever you listen and share Thoughts on the Market with a friend or colleague today.

What separates an embedded finance partnership that changes customer behavior from an integration nobody would miss? In this episode of Tech Talks Daily, I speak with Rory Herriman, Chief Technology Officer and Chief Operations Officer for Zip's US business. Rory works across product, technology, operations, and business strategy, giving him a broad view of what happens after the API connection is complete and real customers begin using the service. Rory challenges a common understanding of embedded finance as placing one financial product inside another company's experience. Customers rarely wake up wanting embedded finance. They want to complete a purchase, manage their money, or solve a practical problem without an unnecessary interruption. The real test is whether the two businesses create something together that neither could provide independently. Rory calls this derived product market fit. Both products may succeed separately, but the combined experience must generate additional value for the customer if the partnership is going to last. Technology integration is only one part of the work. As businesses add customers and partners, they create new customer journeys, compliance obligations, servicing models, governance requirements, and operational processes. Rory argues that this complexity grows exponentially rather than linearly. This changes how technology teams should approach architecture. Instead of creating another custom integration for every opportunity, each partnership should contribute reusable capabilities to a wider platform. APIs, shared services, configuration tools, support processes, and governance models can then serve the growing ecosystem. We also discuss what partnership conversations reveal. Rory sees customer journey discussions during the first meeting as a positive sign. A conversation dominated by revenue division or integration mechanics may indicate that the participants have not established why the customer needs the combined service. His internal test is refreshingly simple. If the company launched the capability and removed it several months later, would the customer notice? If the answer is no, the partnership may have created technical activity without meaningful customer value. AI also enters the discussion. Rory believes AI can move financial services toward adaptive experiences where the product responds to the customer's circumstances. This offers opportunities for personalization and automated servicing, but it also increases the importance of responsible design, governance, customer consent, and clear accountability. For leaders building one-to-many embedded finance models, Rory's advice is to begin with the customer journey, establish alignment on values and service expectations, and build platforms that become stronger with each partnership. Would your customers miss the financial services you are embedding, or are they simply another feature occupying space in the journey? Listen to the episode and share your thoughts with me.

In this episode, Conor and Ben chat about the design of APIs, UIs & algorithms!Link to Episode 298 on WebsiteDiscuss this episode, leave a comment, or ask a question (on GitHub)SocialsADSP: The Podcast: TwitterConor Hoekstra: LinkTree / BioBen Deane: Twitter | BlueSkyShow NotesDate Recorded: 2026-07-22Date Released: 2026-08-07C++Now 2018: Ben Deane “Easy to Use, Hard to Misuse: Declarative Style in C++”NDC Toronto UI talkThe Design of Everyday ThingsAbout Face: The Essentials of Interaction DesignUser Interface Design For ProgrammersJust My TypeThe Elements of Typographic StyleIntro Song InfoMiss You by Sarah Jansen https://soundcloud.com/sarahjansenmusicCreative Commons — Attribution 3.0 Unported — CC BY 3.0Free Download / Stream: http://bit.ly/l-miss-youMusic promoted by Audio Library https://youtu.be/iYYxnasvfx8

What happens when AI search evolves into real-time intelligence for answers, agents, and enterprise workflows?In this episode of XTraw AI, Raghu Banda speaks with Saurabh Sharma, Chief Product Officer at You.com, about the future of AI-powered search, grounded answers, real-time web intelligence, and agentic workflows. The conversation explores how enterprises can move beyond traditional search and standalone AI models toward trustworthy, current, and actionable AI experiences.• How AI is changing search from links to answers, reasoning, and agents• Why grounded answers, citations, freshness, and trust are critical for enterprise AI adoption• How You.com is building real-time web intelligence and APIs to support AI applications, agents, and enterprise workflowsYou can reach @ SaurabhMy LinkedIn @ RaghuWebsite @ XTraw AI

This interview was recorded for GOTO State of the Art in April 2026.https://gotopia.techRead the full transcription of this interview here:https://gotopia.tech/articles/450Abby Bangser - Principal Engineer at Syntasso & Team Topologies AdvocateCharles Humble - Freelance Techie, Podcaster, Editor, Author & ConsultantRESOURCESAbbyhttps://bsky.app/profile/abangser.bsky.socialhttps://twitter.com/a_bangserhttps://github.com/abangserhttps://www.linkedin.com/in/abbybangserhttps://www.syntasso.io/members-area/abby/profileCharleshttps://bsky.app/profile/charleshumble.bsky.socialhttps://linkedin.com/in/charleshumblehttps://mastodon.social/@charleshumblehttps://conissaunce.comLinkshttps://blog.container-solutions.com/paula-kennedy-on-platform-team-responsibilities-patterns-and-anti-patternshttp://sites.libsyn.com/406853/syntasso-coo-paula-kennedy-on-platform-team-responsibilities-patterns-and-anti-patternshttps://www.oreilly.com/library/view/platform-as-a/0642572243777https://github.com/Cloud-Native-Platform-Engineering/cnpe-community/issues/79https://www.kratix.iohttps://leaddev.com/ai/nobody-knows-what-programming-will-look-like-in-two-yearshttps://leaddev.com/ai/shipping-faster-thinking-less-the-ai-code-verification-traphttps://www.cncf.io/blog/2023/11/20/announcing-the-platform-engineering-maturity-modelhttps://platformengineering.com/features/portals-and-pipelines-arent-enough-avoiding-the-platform-facadeDESCRIPTIONAbby Bangser opens with a clear-eyed status report on platform engineering: the concept of centralizing shared capabilities with self-service delivery is well understood, but the execution keeps going wrong in the same way. Organizations move from DevOps to platform engineering, but their platform teams end up becoming the new bottleneck — a centralized group drowning under the weight of the entire organization's requests, which is exactly what DevOps was supposed to fix. Abby traces this to an architectural problem: too many platforms are still built as centralized Terraform machines rather than as a marketplace of composable offerings. Her "platform as a product" test is blunt and useful: has the team ever said "no" to a feature request, or deprecated something?If not, they don't have a product — they have a request queue.The AI dimension is where the conversation gets most urgent. Abby's position is direct: AI agents are the new forcing function for platform maturity. The biggest misconception she wants to dismantle is the persistent equation of platform engineering with infrastructure-as-code: renaming your Terraform team doesn't count. Platform engineering is about building an experience — for human developers and increasingly for AI agents — that is self-service, compliant, and coherent at organizational scale. The Team Topologies model of interaction modes (from high-collaboration to fully automated on-demand APIs) gives a useful health check for where a platform actually sits on that maturity curve.RECOMMENDED BOOKSChankramath, Cheneweth, Oliver & Alvarez • Effective Platform Engineering • https://amzn.to/3OnxN8iGregor Hohpe • Platform Strategy • https://amzn.to/4cxfYdbBlueskyInstagramLinkedInFacebookCHANNEL MEMBERSHIP BONUSJoin this channel to get early access to videos & other perks:https://www.youtube.com/channel/UCs_tLP3AiwYKwdUHpltJPuA/joinLooking for a unique learning experience?Attend the next GOTO conference near you! Get your ticket: gotopia.techSUBSCRIBE TO OUR YOUTUBE CHANNEL - new videos posted daily!

What happens when an enterprise AI agent can retrieve thousands of data points but cannot understand the customer, decision, or business moment in front of it? In this episode of Tech Talks Daily, I welcome back Boris Bialek, Vice President of Industries and Global Field CTO at MongoDB. We examine why the enterprise AI conversation has become more professional as organizations move beyond demonstrations and begin putting agentic systems into production. Boris argues that many companies do not have a shortage of data. Their problem is turning scattered data into information and then into usable knowledge. A bank balance is data. A complete view of a customer's relationship with the bank is information. Recognizing that the customer is currently researching a mortgage and may need assistance within the next 20 seconds is knowledge. This distinction leads to Boris's concept of a knowledge garden. Structured records, unstructured content, live signals, conversations, and business context are organized around a customer or outcome. Different departments can access the parts relevant to their work while AI agents receive the context needed to respond quickly. We also discuss integration debt. Boris recalls one system that required 18 seconds to assemble a customer view and says many enterprises are working with approximately 40 primary data sources. An agent can spend so much time coordinating access across APIs, caches, and applications that the business problem becomes secondary. Trust becomes equally important once an AI agent can act. Boris introduces two measures: the agent confidence score and the business risk score. The first evaluates whether an agent's output appears reliable based on its data, behavior, and context. The second considers the consequences of allowing that decision to proceed automatically. Together, these scores can help organizations decide which actions should pass automatically, which need further machine validation, and which should reach a human reviewer. Boris also explains why data lineage and complete audit trails must be designed into production systems from the beginning. For teams beginning this work, his advice is practical. Choose one business outcome, connect two or three relevant data sources, create a working prototype, and involve business and technical leaders in the same conversation. The goal is to demonstrate how data, context, confidence, risk, and human review work together before expanding the system. Does your organization have an AI data problem, or does it have a knowledge and context problem? Listen to the conversation and share your thoughts with me.

8/5/26Episode SummaryTwo things about AI are true at the same time: it's genuinely powerful when wired into your systems, and getting it there is a lot harder than the demos suggest. In this episode Scott gives you both halves. He walks through Client OS, the tool JadePuma built and runs on client brands — a connector that sits between Shopify, Klaviyo, Notion, GitHub, and Google Drive, holds a detailed brand definition, and routes each task to whichever AI model fits it best by API.The demo covers what's actually running today: automated 404 reports with confidence-scored redirect recommendations, scheduled ADA compliance checks that hand errors to a human expert, brand-compliance scoring on existing pages and emails, sentiment analysis on customer reviews to keep brand language in sync with how customers actually talk, and landing pages built from the theme's existing sections. Coming next: ad platforms, social, Google Analytics, and a connection to the marketing calendar — with the goal of one coordinated campaign across every channel, assembled by AI with human checkpoints built into every skill.Then the honest part. Client OS has taken more than a full-time month from JadePuma's strongest developer and it isn't finished. The connections break, APIs change, and chaining tasks together multiplies the failure points. That math works for an agency amortizing the build across a roster of brands; it usually doesn't work for a single store. Scott's takeaway for store owners: don't build this yourself — ask your agency what they're building, how they're using AI on your brand, and what they'd never let it touch.Show LinksLeave a review - https://ratethispodcast.com/solutionsVideo & Transcripthttps://jadepuma.com/blogs/the-shopify-solutions-podcast/episode-191-what-ai-can-do-for-your-shopify-brand-today

The world of IT is filled with technical qualifications in all manner of disciplines. They've become an expected baseline for knowledge in broad fields like networking and security, as well as narrower, vendor-specific knowledge and skills. But what happens as IT leaders move up to more senior roles? Steve Clarke, cofounder and director of Freeman Clarke joins host Eric Hanselman to talk about the challenges that senior IT professionals have in establishing their bona fides as technical business people. The role of a CIO, CTO, CISO or CDO requires technical depth, but the more important part is the ability to integrate that knowledge with business operations. It's not an easy transition to undertake and even more complicated to identify.  Discussions that are taking place in the 451 Research 451 Alliance community brought up the idea of a business level certification process and how it would have to characterize that unique knowledge set. The Standard is an approach that Steve and his colleagues began to address this issue. The 451 Alliance membership is free and open to all qualified IT professionals and business leaders who want to contribute to industry-leading research. Members take anonymous online surveys in exchange for access to research results that power their IT strategy.   More S&P Global Content: Join the 451 Alliance The Standard Certification The Problem with the Status Quo For S&P Global Subscribers: US tech spending intent dips in Q2, but projects to grow in Q3 – Tech Demand Indicator Highlights from Q2 2026 Beyond connectivity: A CIO's road map from network APIs to business impact Enterprises cautiously optimistic for AI amid data management challenges Host/Author: Eric Hanselman Guest: Steve Clarke, cofounder and director of Freeman Clarke and Tech Leaders Connect Producer/Editor: Dylan Scheible Published With Assistance From: Feranmi Adeoshun and Sophie Carr

Higher Ed AV PodcastEpisode 362Practical AI for AV with Dan LitvinAI may be the biggest buzzword in technology, but what does it actually mean for audiovisual professionals and the spaces they support? In this episode of the Higher Ed AV Podcast, Joe Way welcomes Dan Litvin, co-founder and president of PureTek Group and co-founder of unRAVL, for a practical conversation about moving AI beyond the hype and into real-world AV workflows.Dan explains how AI-powered control can make technology feel more human by allowing systems to adapt to the people using them, not forcing users to adapt to rigid programming. They explore personalized room experiences, natural-language control, open APIs, proactive system monitoring, automated audio and camera operation, and the convergence of AV with lighting, HVAC, security, scheduling, and other campus platforms.The conversation also challenges higher education AV professionals to think beyond individual rooms and traditional support models. By using AI to create smarter, more flexible spaces, AV teams can improve the user experience, reduce operating costs, support sustainability, generate revenue, and demonstrate measurable institutional value. The future of AV leadership will not simply be about managing technology, it will be about using technology to solve the larger problems facing the institution.Also in this episode:Why “making technology more human” does not mean removing humansHow AI can personalize one room for dozens of different users and use casesUsing AI as a translation layer between traditionally disconnected systemsThe importance of open APIs and interoperable technology ecosystemsMoving from reactive support calls to proactive system managementAI-assisted audio leveling, camera tracking, and room controlTurning campus AV spaces into revenue-generating assetsWhy AV professionals must become managers, strategists, and business leadersDan's journey from engineering and yoga instruction to human-centered technologyWhy AI should eliminate tedious work and give people more time for what mattersConnect with Dan Litvin:LinkedIn: https://www.linkedin.com/in/dan-litvin-27959247/PureTek Group: https://puretekgroup.comunRAVL AI: https://unravl.net 

PHP Alive and Kicking – August 4, 2026 Hosts: Chris Miller (standing in solo while Mike is on holiday) Mike escaped to a beach so Chris hosts alone with special guest Derick Rethans. Expect date-time philosophy, PHP 8.6 goodies, Xdebug secrets, deprecation drama, and a heated debate about how to queue at a pub. From University Websites to Committing to PHP Derick traces his PHP origins back to his university days, building websites that needed database connectivity. ASP.NET was an option, but nobody wanted to pay for it — Linux was cool, and PHP was easy to start using. That combination hooked him early. His very first accepted contribution came from a practical need: everyone wanted centered Shockwave Flash films on their pages, back before CSS even existed. Centering meant knowing the width of a file, and while getimagesize() already existed (and still works today), it didn’t understand Flash. Derick added that support. From there he fell into bug triage, spending several years working through the issue tracker alongside Jani Taskinen (whose “Sniper” name still shows up in commit logs). For a long time it was just the two of them reproducing and fixing bugs — an important task that gets neglected easily, which is exactly how backlogs of a thousand issues and pull requests build up. The Eternal Problem of Date and Time Chris and Derick dig into why date-time bugs are so hard to triage: date algorithms are genuinely complicated, and it takes real time to figure out whether a report is a user’s mistake or an actual PHP bug. Crashes are easy to reproduce; subtle date arithmetic is not. The core philosophical question is “what does adding a month even mean?” If it’s January 31st and you add a month, PHP historically just bumps the month number, producing February 31st, which overflows to March 2nd or 3rd. But most people who add a month really just want the next month and don’t care about day counts — so “next month” is conceptually distinct from “add a month.” Derick has been brainstorming better interfaces so people can’t make these mistakes, though it’s a lot of work. They touch on Carbon as a unified interface people understand, but agree that making Carbon the language standard isn’t the right move. At the language level you need something that works for everybody and ideally doesn’t break existing code. What’s Coming in PHP 8.6 The first new time class is landing: a Duration object representing seconds and nanoseconds within a specific range, living in the Time namespace. It’s a trial run for new APIs and immediately useful — for example, passing a duration to sleep or to timeouts in the new polling API, rather than juggling raw integers. Another highlight is partial function application. With the pipe operator from 8.5, piping into anything that takes more than one argument meant wrapping it in a closure (and extra parentheses because of how the parser works). Partials let you pre-fill some arguments so the resulting callable accepts just one — perfect for pipelines. There’s also TLS session resumption support for streams, which lets OpenSSL shortcut the expensive session setup on repeated TLS connections. Bigger stream-layer work by Jakub Zelenka of the PHP Foundation is underway too, but that’s a huge job on old code and won’t make 8.6 — Chris and Derick both note the pain of keeping a year-and-a-half-long project in sync with the master branch. Xdebug in 8.6 and the Xdebug Cloud Xdebug’s recent headline feature (shipped with 8.5) is native path mapping, which lets you configure remote-to-local path mappings inside Xdebug itself instead of your IDE — handy when your code runs in Docker or on a dev machine. Feedback since release is driving quality-of-life improvements. For 8.6, relatively little is broken, though a change to how PHP reports line numbers needs untangling so debugging stops on the correct lines. Derick also has a new code-coverage implementation that gives better, more correct results at roughly a 10% speed cost. He’d rather ship correct output than fast output — and points out that PHPCov may be quicker but skips path and branch coverage entirely. The pair discuss how branch coverage doubles your test count at every branching point, and how Xdebug has a hard limit (around 64K) to avoid running out of memory on pathological code. Finally, Chris asks about Xdebug Cloud — a paid service that enables debugging where networking otherwise wouldn’t, such as remote teams sharing a single development machine (an SSH tunnel only lets one person listen on the debug port). It’s £15/month or £150/year for an individual, priced similarly to Packagist’s private packages, and helps fund Derick’s ongoing Xdebug work. He plans to do more marketing about it this year. Deprecations, Consensus, and Two-Thirds Votes Every year PHP collects a big list of deprecations — around 20 to 25 this cycle — voted on separately. Derick is more reluctant than most to approve them, arguing many offer no clear user benefit. His examples: deprecating metaphone() (whose suggested Composer replacement ironically relies on the internal function) and the dechunk filter (which can’t really be removed because HTTP streams use it internally). The list() deprecation is another sore point — Chris uses it, and the vote is tied 21–21 with one abstention, so it won’t pass under the two-thirds rule. Derick reminds everyone that “yes” needs to be twice “no.” Returning from a finally block is another candidate that makes little sense to use but hurts nothing to keep. Chris and Derick reflect on how voting was originally meant to confirm consensus reached on the mailing list, not to be the decision itself. A 50%+1 result isn’t consensus; two-thirds is the compromise, even if true 75–80% agreement would be ideal. Joe notes many deprecations are groundwork to reserve keywords for future RFCs — which raises the perennial question of how you measure real-world usage across Composer’s dependency chains. Infrastructure, the Website, and the Next Generation The PHP infrastructure has been converted to an Ansible-based repository over the last year and a half (largely by Derick, with help from Joe), replacing a wild mix of individually maintained servers — one was still running FreeBSD 4.3. They’ve also migrated to a new CDN, which brought surprises: a default one-month cache, ignoring the query string in cache keys, and even dark/light mode being an organization-wide setting rather than per-user. On the website side, everything currently lives in a single root directory of 1999-vintage code, and there’s a long-pending patch to restructure it. The PHP Foundation’s ambassadors program — with speaking, marketing, and research streams — aims to make the very techy php.net homepage speak to CTOs and newcomers about who uses PHP and why it’s a thriving project (only 31-ish years old, versus C’s 60). The conversation closes on caretakers and the next generation of contributors. Triage, they stress, doesn’t require deep internals knowledge — just reading reports to see if they still make sense is hugely valuable. Security and infrastructure access necessarily stay gatekept for trust reasons, and both agree a bus factor of one (or even two) isn’t good enough. Chris admits he’s struggling to name people who are both nice and know infra. Links from the show: Join us live in Discord — discord.phparch.com Watch live on YouTube — youtube.com/phparch PHP Tek Conference 2027 — Call for Papers now open PHP Arch Swag Store — store.phparch.com Xdebug — and Xdebug Cloud for shared debugging Magazine discount code ALIVE3 — three free months on an annual digital subscription Host: Chris Miller X: @ccmiller2018 Mastodon: @miller@phpc.social Bluesky: @ccmiller2018.bsky.social PHPArch.me: @miller Mike Page Mastodon: @MikePageDev@phpc.social PHPArch.me: @mikepagedev Streams: Youtube Channel Twitch Connect & Hire PHP Architect Website Twitter/X Mastodon Hire PHP Developers Looking to hire PHP developers? Email support@phparch.com – the team is available for consulting, infrastructure work, Ansible playbooks, and code review. Partner This podcast is made a little better thanks to our partners Displace Infrastructure Management, Simplified Automate Kubernetes deployments across any cloud provider or bare metal with a single command. Deploy, manage, and scale your infrastructure with ease. https://displace.tech/ OurCVEs Your security posture, on autopilot with OurCVEs CodeRabbit Cut code review time & bugs in half instantly with CodeRabbit. PHP Architect Consulting Your PHP codebase deserves a partner, not a contractor PHP Architect provides long-term technical partnerships for organizations that need senior-level PHP expertise that you can depend on. https://www.phparch.com/consulting/ Music Provided by Epidemic Sound https://www.epidemicsound.com/ Join Us Live Next Week Youtube Channel Got feedback? Join us on Discord at discord.phparch.com The post PHP Alive and Kicking 2026.08.04 appeared first on PHP Architect.

In this UC Today interview, host Kristian McCann sits down with Kunal Shukla, GM and Head of Network APIs at Vonage, and Chung H. Lee, AVP of Product Innovation at AT&T Business, to unpack why traditional authentication methods are breaking down—and what comes next.From the hidden cost of fraud to the growing limitations of SMS-based verification, the discussion reveals why enterprises must rethink how they balance security with seamless user experiences. More importantly, it introduces a new, network-led approach that leverages telco intelligence to authenticate users silently—without adding friction.This isn't just theory. Vonage and AT&T break down how network APIs are transforming authentication and reducing friction across the customer journey.Key points include:Why a $10 trillion cybercrime economy is forcing businesses to rethink authentication strategiesHow SMS OTP creates friction—and opens the door to phishing, interception, and poor UXWhat “silent authentication” looks like using real-time network intelligence and telco dataHow Vonage, AT&T, and partners are building an ecosystem to deliver secure, seamless digital trustNext Steps:Stay ahead of fraud trends and join the UC Today LinkedIn community at linkedin.com/groups/9551264 and subscribe to the newsletter at uctoday.com/register.

In this UC Today roundtable, host Kristian McCann brings together leading security experts to confront the escalating risks of agentic AI—autonomous systems that make decisions, take actions, and roam across enterprise tech stacks often without human oversight. You'll hear from Irina Tsukerman, President at Scarab Rising; Shlomi Beer, Co-Founder & CEO at ImpersonAlly; and Roey Eliyahu, Co-Founder & CEO at Salt Security. Together, they reveal how rapid adoption in customer support, sales, and finance is outstripping governance, enabling silent breaches via prompt injections, privilege accumulation, and API exploits that blend into normal workflows.Expect straight talk on the productivity boom versus harsh realities—like 80% of companies facing unintended agent actions such as data leaks or unauthorized access—and proven strategies for observability, dynamic guardrails, and non-human identity governance that safeguard UC&C environments without killing innovation.Talking points include:Why agentic AI supercharges insider threats and erodes perimeters, exposing "internal" APIs to external manipulation even within legitimate agent permissions.Governance blind spots in high-stakes sectors like retail and airlines (e.g., fraudulent refunds, customer data spills), demanding end-to-end visibility from LLMs to backend actions.Runtime defenses against self-learning exploits, deceptive "sleeper" code, and endless trial-and-error attacks that humans resist but agents can't.Zero-trust models, NHI in IAM, and regulation readiness to align agent efficiency with compliance and CX protection.Next StepsShare this roundtable with your CISOs, IT leaders, and business units to map agent permissions and API exposures right now.Evaluate IAM for non-human identities and pilot tools like Salt Security for holistic agent discovery.Subscribe to UC Today for more expert roundtables on AI risks, UC&C security, and enterprise compliance.

Wie entwickelt man Software, wenn ein Breakpoint das Systemverhalten verfälscht, Speicher knapp ist und ein Update nicht einfach per Knopfdruck ausgerollt werden kann? Genau in diese Welt tauchen wir in dieser Episode ein. Wir sprechen über Real Time Operating Systems, Embedded Systems, Mikrocontroller und die Frage, was Echtzeit in der Praxis wirklich bedeutet.Mit Roland Lezuo schauen wir hinter die Kulissen moderner Embedded Entwicklung. Es geht um RTOS, Linux mit Echtzeitfähigkeit, Interrupts, harte und weiche Deadlines, PCB Design, Firmware in C, Treiber, Debugging mit Oszilloskop, Tracing auf Hardware-Ebene und die Realität von Testing und Continuous Integration im Embedded Umfeld. Außerdem klären wir, warum ein smartes Fernglas ein ziemlich gutes Beispiel für anspruchsvolle Echtzeitsoftware ist und weshalb Hardwareprojekte oft ganz andere Kompromisse verlangen als Cloud-Software oder klassische Backend Entwicklung.Zum Schluss sprechen wir über Updates und Over the Air-Update-Szenarien, den Cyber Resilience Act und darüber, wie du selbst in Embedded Software einsteigen kannst, ohne gleich ein Labor voller Spezialhardware aufzubauen. Wenn du wissen willst, warum C, Echtzeitbetriebssysteme und Embedded Linux noch lange nicht von gestern sind, dann ist diese Folge genau dein Ding.Bonus: Print-Debugging mit Oszilloskop ist wirklich so wild, wie es klingt.Unsere aktuellen Werbepartner findest du auf https://engineeringkiosk.dev/partnersDas schnelle Feedback zur Episode:

Send a text to Melissa and she'll answer it on the next episode. Summer sneaks up with more than sun-kissed skin—sometimes it brings pounding headaches, nausea, and a wave of fatigue that won't quit. We unpack the real difference between sunburn and heat exhaustion, explain why they don't always arrive together, and share a clear plan to recognize danger early and respond with confidence. Along the way, we trade personal stories—including a sudden heat crash at a theme park—and translate those moments into practical steps you can use anywhere: shade, smart hydration, and the right remedies at the right time.We break down sunburn care first, from recognizing bright-red, throbbing skin to managing blistering pain. You'll learn when Belladonna, Cantharis, Apis, Urtica urens, and Calendula fit, and how sensations like burning, stinging, and throbbing guide your choice. Then we move into whole-body heat stress: spotting dizziness, weakness, confusion, and cramps before they escalate. Gelsemium, Glonoinum, Natrum carbonicum, Veratrum album, and Carbo vegetabilis each play a role depending on the pattern—whether it's a heat-induced headache, chronic sun intolerance, collapse with cold sweat, or that depleted “reviver” moment when breath feels shallow and energy is gone.We also talk prevention that actually works. Electrolytes matter more than most people think—sodium, potassium, and magnesium help your body cool itself efficiently. We share what to look for in mixes, why small sips beat chugging plain water, and how diet choices and gentle morning sun can improve tolerance over time. Practical tools like cooling towels, pacing activity, and stepping out at the first hint of pinkness keep days at the beach, ball games, or theme parks safe and fun.Listen now to build a lightweight summer plan you can trust: know the signs, match the remedy, and cool the body without panic. If this helped, follow the show, share it with a friend who lives outdoors in summer, and leave a quick review—what's the first sign you notice when heat starts to win?You may also gain Access to my Fullscript dispensary and save 30% by going to: https://us.fullscript.com/welcome/mcrenshaw FIND ME!

In this installment of their recurring roundtable, Jacob, Ari, and Rob dig into the accelerating Chinese open-source model race, debating whether Kimi K3 has actually closed the gap with the US frontier or just looks like it has, and whether distillation fully explains China's progress. That opens into the messier fight over open-weight models generally, with the group dissecting the backlash against Anthropic's stance and asking whether powerful open models are inherently dangerous or actually a necessary part of collective defense. They connect this directly to the Fable ban episode as a possible preview of a coming frontier-model licensing regime, then turn to the OpenAI and Hugging Face security incident as a landmark, publicly legible moment of autonomous AI causing real-world harm. From there the conversation shifts to business strategy: how much value are companies actually handing over to frontier labs by building on their APIs, and what does the Grok and Cursor data advantage really tell us about the future value of real-world usage data versus purchased training data. They close with a rapid-fire tour of the news cycle, covering SSI's mysterious $5B raise, the rumored Stripe and OpenRouter acquisition, Google's continued underperformance despite structural advantages, and OpenAI's leadership speculation, plus a reflection on venture's swing back toward deep tech and physical infrastructure.   (0:00) Intro (2:41) China's Open Source Models Catch Up (7:42) Does Distillation Explain China's Rise? (13:54) The Geopolitical Risk of Chinese AI Models (20:28) Should the Government Restrict Open Models? (22:06) What Are the Labs Really Learning From You? (29:22) Future of Government Regulation (40:50) The OpenAI-Hugging Face Hack (47:59) Grok, Cursor, and the Value of Real Data (56:52) SSI and OpenRouter (1:02:52) Venture Capital's Return to Deep Tech (1:05:16) Quickfire With your host: @jacobeffron - Managing Director at Redpoint

Watch the full episode on YouTube:We first covered Baseten last year when DeepSeek mania was at peak hype. Now they have raised a monster $13B round and become one of the new cohort of AI Infra decacorns that are (with Nvidia, Intel, and the semis complex) chief beneficiaries of the Inference Inflection. We return to Baseten at the peak of the 2026 edition of Open Weights debate. Ali has published a viral breakdown of Kimi K3:And since you last saw him, Philip has spoken at AI Engineer and written the definitive book on Inference Engineering spotted all over SF:Three years ago, inference engineering barely existed as a category.Today, it is one of the most critical disciplines in AI. Inference engineering inherently tackles a different question than standard model training: “How do you turn those weights from training into a product that is fast, reliable, and affordable at scale?” Focusing on these creates an entirely new optimization problem.In one recent GLM-5.2 experiment, quantizing more of the model actually preserved its benchmark quality while increasing throughput by 20%, because the errors introduced in different layers could cancel each other out.Inference is no longer just the final step after training. It is becoming its own engineering discipline, with its own research problems, infrastructure, and increasingly specialized roles.In this episode, Baseten's Philip Kiely and Ali Taha join swyx and Vibhu to explain what actually happens after a new open model is released and what it takes to turn “we generated a token” into a fast, reliable, production-ready API.We go deep on cache-aware routing, disaggregated prefill and decode, quantization, speculative decoding, KV-cache movement, model parallelism, GPU kernels, and the race to make frontier models up to 10× faster. Philip and Ali explain why inference optimizations can still produce gains of 20%, 100%, or even 200%; how quantization errors can cancel one another out; why identical weights can behave differently across clusters; and how Baseten grafted a Kimi vision encoder onto GLM-5.2 without changing the underlying language model.The conversation then expands beyond LLMs into NVIDIA Dynamo, mega kernels, Rubin, AI-specific chips, local inference, video generation, diffusion versus autoregressive models, and the enormous compute barrier to generating coherent long-form video. Finally, we explore the convergence of training and inference, continual learning through persistent KV cache, and the emerging loop where models help optimize the infrastructure that runs them.We discuss:* What happens when a 200,000-token request enters an inference system* Cache-aware routing and reusing previously computed KV cache* Why prefill and decode are increasingly handled by different GPUs* When dedicated deployments become cheaper and more reliable than shared APIs* How speculative decoding uses a smaller model to accelerate a larger one* Tool calling, structured outputs, and what LLMs actually do* What it takes to support a new open model on day zero* Grafting Kimi's vision encoder onto GLM-5.2* Retrofitting inefficient model layers with components from other architectures* Why models sometimes collapse into repeating the same token* How hardware, kernels, and race conditions create nondeterministic failures* Preserving model fidelity while making inference faster* How quantization errors can cancel each other out* Why inference optimizations still deliver gains of 20%, 100%, and 200%* How optimized serving can make a model up to 10× faster* NVIDIA Dynamo, KV-aware routing, and distributed model serving* Speculative decoding the speculative decoder* Why local AI is about making models less dumb while data-center AI is about making them less slow* Tensor, expert, and pipeline parallelism across GPUs* Hardware-aware model design, auto-tuning, and the case against mega kernels* Rubin and why inference is becoming a systems problem* Whether modern GPUs are evolving into programmable AI ASICs* Why enormous models like Kimi K3 require GB300-class hardware* Why open-source video generation still trails Veo, Kling, and other closed models* The quadratic attention bottleneck behind long-form AI video* Autoregressive video, real-time generation, and compounding quality drift* Why future video systems may combine autoregressive and diffusion architectures* Training for inference and inference for training* Continuous post-training, deployment, evaluation, and improvement loops* How GLM-5.2 helped optimize the kernels serving GLM-5.2 itself* Why faster networking could unlock dramatically faster decoding* Continual learning, KV-cache compaction, and persistent model memoryShow Notes* How to build a day-0 API for Kimi K3* 22580: From GPT2 to Kimi3, ExplainedPhilip Kiely* LinkedIn: https://www.linkedin.com/in/philipkiely* X: https://x.com/philipkiely* Inference Engineering: https://www.baseten.co/inference-engineering/Ali Taha* LinkedIn: https://www.linkedin.com/in/aliestaha/* X: https://x.com/waterloointernTimestamps00:00:00 Introduction and the 200K-Token Prompt00:03:18 Dedicated Deployments, Speculative Decoding, and Tool Calling00:11:26 Launching Production-Ready Open Models00:19:06 Model Retrofits, Failure Modes, and Nondeterminism00:28:22 Quantization and Canceling Errors00:32:15 The Race to 10× Faster Inference00:40:48 Dynamo, Speculation, and Local vs. Data-Center AI00:50:18 Model Parallelism, Auto-Tuning, and Mega Kernels01:00:55 Rubin, GPUs vs. ASICs, and Custom AI Chips01:10:03 Giant Models and the Limits of GPU Memory01:12:42 AI Video, Quadratic Attention, and Autoregressive Generation01:21:47 Audio, Images, and Diffusion Models01:27:32 Training, Self-Optimizing Models, and Continual Learning01:40:06 Closing ThoughtsTranscriptIntroduction: Baseten, Waterloo Intern, and Inference EngineeringSwyx [00:00:00]: Okay, we're here in the studio with Philip, old friend from Inference Engineering, the book, as well as Baseten and everything that you've done, you and I have done before, as well as Ali. Welcome.Ali [00:00:15]: Pleasure to meet you.Swyx [00:00:15]: Waterloo intern.Ali [00:00:16]: Waterloo intern, always.Swyx [00:00:17]: When did you get “Waterloo intern” as a handle?Ali [00:00:19]: As a handle? Oh.Ali [00:00:20]: I think the rebranding happened mid-March. When I saw it was open, I was like, “I have to take it. Up for grabs.”Philip [00:00:26]: The problem is that Ali is really good at his job and is not gonna be an intern much longer.Philip [00:00:30]: So we have to figure out who's gonna get the handle.Ali [00:00:33]: Well, I'll pass the torch over to the next intern.Swyx [00:00:34]: Oh, okay. It can be, like, you just pass it to another Waterloo grad.Ali [00:00:37]: To another Waterloo intern. No, bruh.Philip [00:00:39]: Yeah.Ali [00:00:39]: Intern.Swyx [00:00:40]: Intern, yeah.Ali [00:00:40]: And no.Philip [00:00:41]: You gotta get an intern from Waterloo.Ali [00:00:42]: Yeah, I've gotta get an intern from Waterloo.Swyx [00:00:44]: Right.Ali [00:00:44]: But they have to follow the path.Swyx [00:00:45]: Oh, it could, but it could come from Baseten, so it's like whoever Baseten gets from Waterloo.Ali [00:00:48]: Right.Swyx [00:00:49]: Has the title of Waterloo.Ali [00:00:50]: It stays in the ecosystem.Philip [00:00:51]: Exactly.Ali [00:00:52]: Halfway through the internship, you either get it or you're out.Philip [00:00:55]: You should also do, like, a big graduation ceremony where you change the handle.Ali [00:00:59]: Just say it.Philip [00:00:59]: For everybody.Swyx [00:01:00]: You guys are good at ceremonies, clearly. We had a nice launch of the book, very successful. But before we get into all that, I wanna start off with a fun question for you. Okay, you're an expert inference engineer. What happens when I send a long query, say two hundred thousand tokens into Baseten's inference? What's the process of query through GPU model routing, balancing, all that? What is all the stuff that we don't think about?Long Context Requests, KV Cache, and Cache-Aware RoutingPhilip [00:01:26]: With a long query specifically, the first thing that I'm gonna ask is, “Have you sent me this query before, or at least part of it?” and I really hope you have, because it's gonna be a lot easier for me and a lot cheaper for you. So the first thing that we're gonna look at is some cache-aware routing, where we're going to see, we probably have a number of instances, a number of replicas up serving whatever model you're hitting. We want to send this one to something with, number one, available prefill workers, and number two, ideally some cached input already there so that we can skip prefill on at least part of these two hundred thousand tokens. If you're doing two hundred thousand tokens, it's probably coding or a multi-turn agent or something where you would expect to have that cached. If you don't, we're gonna have to send it to a prefill worker. We've at least on certain models disaggregated prefill and decode, so you're going to have one set of GPUs that's solely going to process the input, create the KV cache, and get you your first token, and then that's going to be passed over to a separate set of GPUs, which is going to run decode. We're going to iteratively make those tokens. We're probably going to have some speculator model in front of that. I'm going to assume that you're doing coding, and because of that, our speculator model, which assumes you're doing coding, is gonna have a high draft token acceptance rate. If I'm wrong and you're asking me to summarize every Harry Potter book, it's gonna be slower. And then we stream that output to you and account for it, charge you, a couple of pennies and say, “Hey, would you like to send another one?”Swyx [00:03:04]: Except Baseten doesn't charge by pennies.Philip [00:03:07]: Well, yeah, we charge. I'm assuming that we're talking about the public model APIs. If you are setting up a dedicated deployment, then yeah, it's not pennies.Public APIs vs. Dedicated DeploymentsSwyx [00:03:18]: Yeah, one of the key differentiators when I was talking with Baseten initially was that people who want very high volume just need to rent by the box, ‘cause then it's up to you to figure out how to saturate the box.Ali [00:03:31]: And more often than not, it's, like, way cheaper if you're pushing, like, millions of tokens per hour, if you just pay per hour instead of pay per token.Philip [00:03:37]: Yeah, they do. I think that we've increasingly seen a lot of demand for the pay per token APIs, just because everyone wants to try open models, and then once they find a use case that's really sticky, then they move over to dedicated.Swyx [00:03:51]: Is there a best practice on when it's time to swap over?Philip [00:03:54]: Couple reasons. Yeah, reliability, that's a big one, right?Ali [00:03:57]: Like, if they have a very specific use case, they want you to train something specifically for them, like they want their own spec dec, for instance, for their own traffic.Swyx [00:04:04]: Spec dec is speculative decoding.Speculative Decoding and Custom SpeculatorsAli [00:04:05]: Speculative decoding, yeah.Swyx [00:04:07]: You have to explain.Ali [00:04:07]: Sorry. Like, speculative decoding is like, if you have a huge model, right? And so the model is going to be generating one token at a time every single turn, every single forward pass. So we attach, like, this little, like, parasite, like this layer that goes on top of the model, and this model just has to predict. It does three very fast autoregressive forward passes, and it will predict, like, three certain tokens, and then you do one forward stage over the entire original model in order to see if those predictions were correct or not, and then you accept them or you reject them. Now, this draft model is traffic specific, so if you, like, Philip said, if you're summarizing Harry Potter books, I can train exclusively that draft model on Harry Potter books, and I can guarantee you that I'm gonna accept the three tokens every single time. And so with that case, I increase your decode speed. I wouldn't be able to provide this to you if you're a shared endpointSwyx [00:04:53]: YeahAli [00:04:53]: ‘cause I have no idea if you're doing Harry Potter, if you're doing coding, if you're doing English. We don't know. Also, there was a thing in the book that mentioned that if they really cared about a specific threshold, chapter four, I think. Do you remember that?Philip [00:05:06]: Yeah. The things that you can do is you can set a specific, like, batch sizing, a specific, like, parallelism strategy if you're trying to optimize for, like, throughput versus latency. You can. Maybe a NVFP4 quant doesn't pass your benchmarks and you wanna run a model at higher precision, you could do that. There's just a bunch of reasons why you might wanna have your own endpoint and the biggest one, of course, just being, like, you don't have to deal with someone else doing a hundred million tokens of benchmarking traffic at the endpoint when you happen to be trying to serve your users.Swyx [00:05:40]: Yeah. I think one thing that is. That is a classic journey. Like, it's people is asking the, what happens when you type Google into the browser. Tool calling, is that just, you're generating JSON or is there more complication beyond that?Tool Calling, JSON, and Structured OutputsAli [00:05:58]: Certain customers that we have, they have their own post-trained models, and so they demand a tool calling that's not just, like parse a file or go find the weather. It's something that's very specific and you have to do post-training on this. And if the post-training on the model is not good or if the quantization after the post-training to get the inference to be fast, the model will struggle reading the JSON file and reading the tool calling. But it doesn't require its own like sandbox. It's not like it's going to use that tool calling to like escape a sandbox or like it doesn't have to be contained. It can just be a normal dedicated deployment. The challenge with tool calling more and more seems to be that the companies want certain tool calling which is a very sensitive thing to train. And because you're dealing with all of the JSON outputs, if it doesn't like close the end of the request in a very certain manner, you end up with a model that did the tool calling and like the thinking and so as a result of that, it didn't see the result and just hallucinated the result as it decoded. That seems to be the most challenging thing with tool calling, not really the sandboxes model.Philip [00:06:56]: Yeah, that's a challenge on the training side and then on the inference side, there's work that you can do to scope the possible output. So we published this at this point close to two years ago, the solution to this problem which is you make a state machine and you use that to constrain the output to a specific format. So this is the structured output problem. If you remember backSwyx [00:07:27]: Yeah, the specific grammar is,Philip [00:07:29]: Yeah, exactlySwyx [00:07:30]: GML had this thing.Philip [00:07:31]: Yeah. So it's like the old-school “make sure this is only JSON”, return only JSON orSwyx [00:07:38]: YeahPhilip [00:07:38]: Grandma's gonna die type of prompts.Swyx [00:07:39]: Is it BNF grammar? At some point OpenAI had released a thing that was like, yeah, if you want to constrain your output, write BNF grammar, back as NOR.Philip [00:07:47]: In our inference system, it's just a specified output format. And you get the guarantee that your output's gonna be structured along that format. And so applying that to tool calls can like help cut down on. You can still call the wrong tool or call no tool. It doesn't solve the certainty problem but it at least solves the output structuring problemSwyx [00:08:10]: YeahPhilip [00:08:10]: Within tool calls.Swyx [00:08:12]: And MCP is just another form of tool, right.Philip [00:08:14]: Yeah, exactly.Swyx [00:08:15]: As far as there's no special thing there.Philip [00:08:16]: The thing I'm always like explaining to people is the LLM is not capable of doing anything. It's only capable of making suggestions of what to do and then if those suggestions are formatted in a certain way and applied to a system that knows what to do with them, then an action occurs.Swyx [00:08:32]: Yeah. Part of the fun stuff is, this is solved outside of tool calling too. Like in an agent loop if the output is not correct or you're right, like reasoning, tool calling was done in the reasoning trace, just be like, “Oh, I don't know what to do. Let me just try again.” And it might get there after a few tries. And on your point of training, sometimes this is harder in smaller models, so you don't have the same exact quality outputAli [00:08:56]: Right.Swyx [00:08:57]: When you just swap from a big model, right?Ali [00:08:59]: Yeah. I will say that, before, I think we need to go back to inference engineering proper.Ali [00:09:04]: But, I had expected that something would replace JSON because it's hard to stream JSON ‘cause JSON must be complete and you must have open and close brackets and everything. So it's hard to parse something or validate something while it's being streamed. So people invented all sorts of things that are like, I forget the name of some of these alternatives, but it's something like TOML, something like YAML. But JSON seems to be dominant still.Philip [00:09:30]: The JSON outputs aren't that long, right? Like you could have a long-- ‘cause tool calls also contain the arguments in them and perhaps for a certain tool you might pass like a very long argument. But my impression of the median tool call is that it's a relatively small number of tokens, right? So I would expect that speculators are generally fairly good at something as formatted as JSON. And so you would have like a pretty fast decode step there and that the streaming wouldn't be as valuable, but maybe I'm wrong about that.Ali [00:10:02]: I think you're also bounded by the software or that the model is gonna integrate with if the software is built with JSON for the tool calls or if the company that you'- if your customer says that this is how our software works and our tools are interfaced with JSON, you can ask them to like, change their software and say like, “Yeah, this is gonna be better for the model.” but like with the right training shouldn't be that much of a difference. Also more profitable if it outputs more tokens probably.Swyx [00:10:25]: Depends on your business model.Swyx [00:10:27]: It really depends. But I will say that, as a writer with like experience a lot with generated output, I do try to move from text to JSON text which is very long JSON, right? Like there's paragraphs in every field because I'm trying to structure it, right?Philip [00:10:44]: Right.Swyx [00:10:44]: I want you to first make factual statements, then make opinions then make bullet point summaries, have dates, have entity references have your sources for references, all these things. Anyway, so these are things that like I think people who really experiment with structural output have to really care about. But, let's, let's recurse up the stack a little bit. Before we started recording, you mentioned something really cool, which is that there's a lot of engineering that-- inference engineering that goes on when a new model provider releases a new model, right? So let's call it GLM-5.2, Kimi K3. I had previously assumed, especially if it's like, well, GLM 5 to 5.1 to GLM-5.2, like that you've supported them before. Is it that much work?What It Takes to Support a New Open ModelAli [00:11:26]: It's a lot of work.Swyx [00:11:28]: Yeah. Okay. So like, a lot of people, all you guys, right whenever a new model launch like, people rush to say like, “Oh, Hugging Face supports this, Fireworks supports this, Spacetime supports this,” and I'm like, “Yeah, of course we support it.” But what goes into that? What goes intoPhilip [00:11:40]: I think it's more than just support it too, right? It benefits the consumer a lot. Like I think it was with Kimi K2.5 or GLM-5.2 the latest, there was an inference war, right? X provider is at 90 tokens a second. The next day we're at 150. The nextSwyx [00:11:55]: I kinda kicked that off with the GLM-5.2.Swyx [00:11:58]: I wrote a Twitter article about. It got like half a million views,Ali [00:12:02]: Based on being numberSwyx [00:12:03]: YeahAli [00:12:04]: Or it's for something else.Swyx [00:12:05]: Yeah. Which,Ali [00:12:06]: Oh my GodSwyx [00:12:07]: Which then got everyone really excited about, hey, how can we, bend tracks a little bit further and,Philip [00:12:14]: There's a difference between support the model, as in I can make a token out of this model, and support a model, as in I have a production-ready API from this model.Philip [00:12:26]: Getting to the point of I can make a token out of this model is not that hard because generally the, open source inference engines, vLLM, SGLang of the world oftentimes even receive weights ahead of time, maintainers do, or the people making the model merge PRs to ensure support. So you generally can, just get it working on the standard open source stack without too much pain in most cases. The challenge is, every inference company is gonna have own proprietary stack. Some open source components, some in-house stuff. And for any arbitrary model, there's going to be some new stuff. Sometimes you get lucky, like K, two five to two six was, like, pretty similar.Quantization, Speculators, and Production ReadinessAli [00:13:16]: Yeah. It was pure continued post-trainingPhilip [00:13:18]: YeahAli [00:13:18]: If I remember correctly.Philip [00:13:19]: Even in those cases, there's still stuff you have to do. You have to redo the quantization work. You're taking the model from. Generally, these models are not released in NVFP4, and we want them to be in NVFP4 for maximum Blackwell compatibility. So we have to perform that quantization, and, calibrate the quantization to make sure that we're not causing any regression in the model's intelligence. And then we also have to train the speculator, as we've talked about. Generally, we have. We have ZDR, zero data retention on our model APIs, so we don't know exactly the traffic that people are sending us, but we know what's popular. We know that coding use cases are popular. We know that agents, agentic use cases are popular. So we can get public data sets that are representative of that traffic and train general speculators. Now, with speculators today, you need to train the speculator using the base model itself because you're getting hidden states out of the model from running inference on these specific prompts, and that is the training data you use to create the speculator. So there's that process which you need the real model weights for. And then there's of course just the process of, standing up all the infrastructure behind it, loading all this stuff, testing it. And then when there's a new model with a newer architecture, I think that, like, the DeepSeek models tend to be the most challenging as they have, like, the most novel architectural stuff going on, model after model. But every new model has something. Kimi K2 had. Oh, sorry, GLM-5.2 hadAli [00:14:53]: Sparse attention.Philip [00:14:54]: Yeah,Ali [00:14:54]: YeahPhilip [00:14:54]: the DSA.Ali [00:14:55]: Right. Which is brought from DeepSeek.Philip [00:14:57]: Yeah. AndAli [00:14:59]: So you can copy-paste then?Philip [00:15:01]: It kindAli [00:15:01]: I don't know how this works.Philip [00:15:02]: So, like we had to, like, build support for that into our runtime. And you're right, like it is really interesting the way that all of these open source labs borrow from each other. For example, like GLM-5.2 doesn't have vision. So something that, Haley, a guy on our team, if we could take a look at this, he, like, grafted the Kimi vision encoder onto GLM-5.2.Retrofitting Vision into GLM-5.2Ali [00:15:27]: We'll be training the projector.Philip [00:15:28]: Exactly. So if you think about, like, the encoder, there's the encoder, which is the part that looks at the image and turns it into latent information, and then there's the projector which likeAli [00:15:38]: You can say latent space. It's okay.Philip [00:15:41]: And then there's the projector that maps it onto, the model itself, and then there's the model weights. You don't wanna mess with the model weights because you run a chance of making the model dumber at something else for the purpose of giving it vision. So instead, Haley started with just a projector, which is only a handful of millions of parameters.Ali [00:16:02]: That would be, yeah.Philip [00:16:02]: Yeah.Ali [00:16:03]: Can you show the training one?Ali [00:16:04]: Like the way it groksPhilip [00:16:05]: YeahAli [00:16:06]: Very interesting.Philip [00:16:06]: And maybeAli [00:16:07]: That right therePhilip [00:16:07]: Maybe Ali, you should take it from here. You've got a betterAli [00:16:10]: Ooh, double the sandPhilip [00:16:11]: Understanding of this than I do.Ali [00:16:11]: Yeah. You can see, like, he. The way he trained this is really cool. At the beginning, he was training it using just like, “Here's a picture of a mountain. Can you describe what's in this mountain?” And that caused it just like the first, learning walls. Like here you can see this all we're trying to teach it is to translate the encoded. Like it's already taken the encoder from Kimi K. It's taken the image. It'Philip [00:16:31]: Yeah. FrozenAli [00:16:31]: FrozenPhilip [00:16:32]: With adapter.Ali [00:16:32]: Exactly.Philip [00:16:33]: Yeah.Ali [00:16:33]: So the brain is frozen and the eyes are frozen. It's just we're tryingPhilip [00:16:37]: AlignAli [00:16:38]: Interconnect between the eye and the brain, right? So the projector. And so you take the tokens and then he's like, “Oh, can you describe what's in this image?” And he's like, “Oh, it's a mountain,” or it's a person or it's a human, whatever the case is. But that didn't cause complete understanding. So he changed it such that every image was associated with a data set of questions. Like, does this image have a white male? Does this image have birds in the top corner? Does this image have a scientist in it? All of that stuff. And it would have to answer questions correctly. And using not just training on describing an image, but being able to answer question, another question, answer over time. Like you can see the grokking, which is like genuinely insane, that retrofitting vision into a large LLM can learn to that extent. And even for images that it doesn't perform well on, for instance, if you ask it a picture of like Stephen Hawking, “Who is this?” Maybe it doesn't get it, but it will say something like, “This is Albert Einstein.” Like it still understandsPhilip [00:17:25]: Close enoughAli [00:17:26]: That this is a scientist who is a man who has, some significant achievements, all that stuff. So that's like really cool.Philip [00:17:32]: Yeah. So, we've covered Hao Tian before, who the author of the LLaVA paper that did this, a while ago. And I think that's very foundational work for anyone who hasn't done vision work before.Ali [00:17:41]: Same with the CLIP and MetaCLIP, where you go from just captioning to building out questionsPhilip [00:17:47]: RightAli [00:17:47]: Off the image and how much better you can get performance.Philip [00:17:50]: Right. Right. Right. Yeah. But what's, what's so exciting about this is if you look at a model like this. Now, this is a little bit more of a research project. It's not. It got to 56% on MMLU Pro, I think. So not quite frontier. But if you're running this model, you haven't suffered any loss on your GLM-5.2 quality. If you don't have an image, it'll just behave exactly the way it used to. And ultimatelyAli [00:18:14]: Which in the inference code you literally do not include the other part, right?Philip [00:18:18]: Yeah. You would just skip the encoder if you don't have an image input.Ali [00:18:22]: Okay.Philip [00:18:22]: Just confirming.Philip [00:18:23]: YeahAli [00:18:23]: Does it affect a lot on the overall inference side? Like you're not adding much, you're adding a very small vision encoder. These are typically likePhilip [00:18:30]: They're super fineAli [00:18:31]: Less than a billion parameters, right?Philip [00:18:32]: Yeah. It's, - There's a little bit less standardization among vision encodersSwyx [00:18:37]: YeahPhilip [00:18:37]: So the support matrix can be a little bit, sparser. But overall, yeah, it's a pretty, it's a pretty minor component of the overall system. And ultimately what you get out of the system is all of a sudden you have Kimi Vision, GLM weights, and DeepSeek attention all in one model.Open Source Model Grafting and Franken-MergesPhilip [00:18:56]: And that's, I think, a lot of the power and beauty of open source, is that you can take all of these different components and combine them together into a system that's better than anyoneSwyx [00:19:05]: YeahPhilip [00:19:05]: Can be individually.Swyx [00:19:06]: People used to say that you would also do Franken-merges where you would take likePhilip [00:19:10]: YeahSwyx [00:19:10]: Layers from each model.Swyx [00:19:11]: Does anyone do that anymore?Ali [00:19:13]: Well, to your point previously when you were mentioning like, the work that goes into supporting a model when it first comes out, like GLM-5.2 or MiniMax M3 or whatever the case is. Sometimes you do have to like, you do have to switch out some things. Like, for instance, the MiniMax M3 head uses full attention, and with full attention you end up with this like insane bottleneck in spec dec ‘cause you're doing auto-regressive token generation for three tokens, and you're doing this like N squared over all of the tokens that are in your sequence. Your KV cache is like very large because it's not sparse, it's not top K. So we find it better to like, okay, we're gonna replace this, we're gonna replace this layer with a layer from another model that's using like GQA, for instance. And then just with the right training, you can get it to have the same acceptance rate. So it is very possible to retrofit layers from other models and very much needed. If a layer is like inefficient, the training just becomes the challenge, like how do you ensure that you train it properly? Which again to your earlier point is like the mesh between training and inference. As in like you need very good training in order to do fast inference. That's like, I feel like more and more becoming true.Swyx [00:20:21]: Yeah. Anything else on the support side when you say like get it to fully production ready?Loop Detection, Race Conditions, and Non-DeterminismPhilip [00:20:26]: Yeah. I think that there's also a question of just, we can test a model to a pretty extensive degree, but we're trying to get it out quickly and then you see a bunch of other people test it and you get interesting results. There was an issue with, GLM briefly where we had some like mode collapses where it would just output the same token over and over again for certain prompts on certain temperatures. Like once you expose an endpoint to the real world, there's going to be, so many more varieties of things given to it that you're able to, discover and patch things. So it's not just a, day zero process, it's then like for the first week, for the first month, if a model remains popular, like how do you both fix bugs and then continue to push the envelope on performance?Ali [00:21:21]: What do you mean you don't want your model outputting S?Swyx [00:21:24]: Is there loop detection on that stuff, by the way? It still happens like quite a lot, which is surprising.Ali [00:21:30]: We have like we, in our endpoint, like if a model was to output the same token like four plus times, we just cut the generation. We say like, “Oh, sorry, this-- Like try again,” or like we will reprocess the request. ‘Cause we know then, like if it, like if, yeah, it's four times the same token, it's probably collapsed.Swyx [00:21:45]: Yeah. Is there a way to opt out in case I really want that?Ali [00:21:48]: You want that?Ali [00:21:50]: I think there's a way that we have to handle it. I'm not exactly certain, but I feel like in certain models, like when they output something like you can imagine, like a table for instance, and so they want, they wanna draw like 12 dashes and 12 dashes. Yeah, I think there's a way for that to happen. I think we only do it on certain tokens. Like we exclude certain special characters.Swyx [00:22:07]: Yeah.Ali [00:22:07]: So we only do it on like certain like S is the most common almost. GLM-5.2Swyx [00:22:11]: OhAli [00:22:11]: And I think it was DSV 4 as well. Like you'd just have like looping issues where like you literallySwyx [00:22:17]: ItAli [00:22:17]: Just have like S.Swyx [00:22:18]: Yeah. Is there a special, something special about S? No, just randomlyAli [00:22:21]: It just seems to be the one token involved.Swyx [00:22:23]: Yeah. And it'Philip [00:22:24]: Is thereSwyx [00:22:24]: And it's only temperature 0Ali [00:22:27]: NoSwyx [00:22:27]: Even at other temperaturesAli [00:22:27]: Even at like 0.9 or whatever, it will still, it will still collapse.Swyx [00:22:30]: That's weird, right?Ali [00:22:30]: It's, it is an inference problem to be honest, like a software problem. Like oftentimes, the image you run will-- like NVIDIA will release an image for instance, and if we will upstream the changes from their latest TensorRT-LLM image into our stack, we'll find that it fixes it. Or oftentimes this will only happen in an inference engine that you're using like SGLang. But if you were to switch to vLLM, that isn't the case. So it seems to be like an extremely like deterministic software issue and not really a model issue. It's not like a weights problem. Like I'- we'll say like, “Oh, it's a problem with the quant. We did PTQ wrong,” right? But that isn't, that doesn't make sense because the same weights used with a different inference engine does not repeat the problem. And sometimes it's, the kernels that are being used in the backend have like these very subtle sometimes race conditions, where if you were to use this model hosted on one cluster, you will never get this problem.Swyx [00:23:19]: Oh my God.Ali [00:23:19]: But if you host it on a different cluster, you will. And the reason is the KV cache transfer from a node to node in that one cluster is using a slower interconnect than the node to node in another cluster. So that exposes the race, whereas in another cluster it doesn't. So then you end up just like, okay, this model is not gonna be hosted on this cluster. We're gonna host it on, another cluster because that cluster exposed that problem. But then it ends up with like, okay, is it the software? Is it the model weights or is it the hardware?Swyx [00:23:42]: There is a thing about this with temperature 0 still not being deterministic, right?Ali [00:23:46]: Right.Swyx [00:23:46]: Mostly because of hardware. Even at temperature 0 same model, you won't always get the same output.Swyx [00:23:52]: Even-- But I'm surprised by the race condition one because, I thought PyTorch was a graph that like guarantees that you at least, execute things in the right order.Ali [00:24:02]: Well, yeah, true. Like I'm not, I'm not saying that there is. Like well, you have things like PTL optimizations where like you can start a kernel before the end of the previous kernel, and that's like ‘cause you want to do that because there'sSwyx [00:24:12]: It's like pipeliningAli [00:24:12]: Expense. Exactly.Swyx [00:24:13]: Yeah.Ali [00:24:13]: But it'- But you don't do it cleanly. Like you overlap a little bit of the execution. No, it is very possible that the kernel itself, like that one block that is supposed to be running in this instance of time, that kernel itself has a race condition. For instance, like a missing barrier. Like often if you're designing a kernel and you want it to make it to be very fast, if you don't test it extensively, you'll, you'll have certain threads access data points from registers before they've been written to by other threadsSwyx [00:24:36]: YeahAli [00:24:36]: For example, because like your barrier is wrong or your synchronization was wrong. But yeah, like the testing itself is very difficult in those like, andSwyx [00:24:42]: And there's no like borrow checkerAli [00:24:45]: What does that mean?Swyx [00:24:46]: Like Rust. Like the. If you're trying to have like memory safety It sounds like a comparable problem.Ali [00:24:52]: Well, yes, but you're working in CUDA, right, NVIDIA GPUs. Like- You just need a higher level language like modular Maybe that's what modular is supposed to do. I don't know.Quantization Quality and Vendor FidelityVibhu [00:25:00]: How do you see keeping quality of the model? So you talked about all these steps of, okay, you gotta do quantization, train your own speculative decoderAli [00:25:07]: RightVibhu [00:25:07]: Run on different hardware. Looking at other model providers, okay, you kicked off a inference speed race on the consumer end. What goes into keeping quality the same across them, right? Sure, you can run benchmarksAli [00:25:22]: YeahVibhu [00:25:22]: But, like, how do you determine how much quantization are there standards? What goes intoPhilip [00:25:27]: There's a few things on quality. Most inference optimizations are lossless. KV caching, for example. You are just recomputing or preventing recomputing the same values. Speculation, of course, if a draft token is wrong, it gets rejected. The main lossy optimization is quantization. And that really comes down to, number one, data format, number two, which parts of the model you choose to quantize, which layers, and number three, like doing a lot of calibration on the quantized weights, to ensure that you're preserving all the outliers. There's other tricks that you can do, though. A big one is long context, ‘cause one thing you asked at, right at the beginning is, “Oh, what's gonna happen if I send a 200,000 token request in?” So with a long input sequence, you need to, store a lot more information. You need to process a lot more tokens. And so even if a model has a context of a certain length, you might, as an inference provider, choose to build an API with a shorter context length, and of course a full length one as well. Because if someone doesn't need the full million token context, for example, you can get them better performance. I don't know if that's exactly like quality of the model. The way that I think about quality is to what degree are we faithfully serving the original model? If you think of a golden implementation of a model that performs exactly the way the model is designed to perform, I think of quality as how close are we getting to that, 100% fidelity of the model.Philip [00:27:13]: You can also, of course, think about quality from the training side and how do you push yourself past 100%. But when I think about purely inference optimizations, it's getting faster while staying as close to that 100% fidelity mark as possible. And certainly our standard internally is that, like you should not be able to tell the difference between our API and a, official API. I think Kimi in particular does a good job of vendor benchmarking hereAli [00:27:41]: YesPhilip [00:27:41]: Where they haveAli [00:27:42]: They released an actual vendor benchmark.Philip [00:27:43]: Exactly, yeah.Ali [00:27:44]: ‘Cause they accused, some people, Amazon? There was some provider that was not doing very well on Kimi's benchmark.Philip [00:27:50]: Yeah.Philip [00:27:51]: So, with Reflect we probablyVibhu [00:27:52]: This was a long time ago, right?Philip [00:27:54]: No.Ali [00:27:54]: Yeah, like threeVibhu [00:27:55]: They alsoAli [00:27:55]: Four, five months agoVibhu [00:27:57]: This also happened with, I don't remember which model, but they pulled out quite a few, and then they started a whole chart about this. It might have beenPhilip [00:28:03]: Kimi Vendor Verifier.Ali [00:28:04]: Yeah.Philip [00:28:05]: Yeah.Ali [00:28:05]: Yeah, ‘cause you, ‘cause you'd be pissed, right? Like if you'Philip [00:28:07]: Yeah.Ali [00:28:07]: If like if I'm a consumer and I'm using like Amazon's endpoint for instance, and I've used Kimi and I'm like, “Oh my God, like this is bad,” I'm not gonna say, “Oh, Amazon quantized the model in a bad way.” I'm gonna say, “Oh, Kimi sucks.” Right?Philip [00:28:17]: Yeah.Ali [00:28:17]: So it seems like that makes sense.Philip [00:28:19]: Yeah, they care. They care.Vibhu [00:28:21]: Justifiably.Ali [00:28:21]: Yeah, justifiably.Vibhu [00:28:22]: This is probably a stupid question, but just checking, has anything improved from main quantization?Philip [00:28:28]: Yeah.Vibhu [00:28:28]: Like, is quantization always strictly worse?Ali [00:28:30]: Well technicallyVibhu [00:28:32]: NoAli [00:28:32]: It's a lossy. QuantizationPhilip [00:28:33]: YeahAli [00:28:33]: Is a lossy, it's a lossy implementation.Philip [00:28:36]: Speed improvesVibhu [00:28:36]: Speed improves.Ali [00:28:37]: It the number, likeVibhu [00:28:38]: No, I' always look for inverse scaling laws.Philip [00:28:40]: Yeah.Ali [00:28:40]: Yeah.Vibhu [00:28:40]: This is something I learned from Noam Brown, where like things that normally act in one direction sometimes do.Philip [00:28:45]: Well, technically when you run a benchmark, because these models are deterministic, sometimes your,Ali [00:28:52]: YeahPhilip [00:28:52]: NVFP4 quant is like, two basis points higher than yourAli [00:28:56]: No, it's noise. It's noise.Philip [00:28:57]: Yeah, exactly. I'm like, yeah, it's, it's within. That's why I always say within margin of error.Philip [00:29:01]: And I stopped saying that because everyone assumes that what is, well, within some margin of error, we're barely inside of that to the worst, so we're saying. But yeah, sometimes it's just like, gives you a higher output score. But like Ali said, that's noise. To my knowledge, you're not necessarily making the results better. You're just trying to, again, like keep your fidelity as close to 100% to the original model.Layer Selection, KL Divergence, and Better QuantizationAli [00:29:27]: There is, to your point, research that we did on MP. I don't know if you are able to pullPhilip [00:29:31]: YeahAli [00:29:32]: A tweet we did. One of our research interns, Joshua, I think it's a tweet on how we have 20% better quantized GLM-5.2 than NVIDIA. Essentially what we found throughout like this month research is, okay, quantization is a lossy. It's. You're compressing the data from, occupying 16 bits to occupying, four bits, for instance. And so you're losing some information, and you're trying to minimize that. And so when I say that I'm gonna quantize the model, my job becomes how do I find the layers that I can quantize, and how to find the layers to not. For instance, with image models, I don't quantize modulation layers, and I don't quantize out projections because those two are. Like out projection is what you see as the user. Modulation is what the model sees or understands. Right, exactly. And so to his paper, do you have the. It doesn't have the. Yeah. It's a long paper. I don't know if I can findVibhu [00:30:25]: If there's a part to search or it's probably in the thread.Ali [00:30:28]: It's probably in the thread.Vibhu [00:30:29]: Yeah.Ali [00:30:29]: But the long and the short is it is very possible that quantizing more of the model makes the results. Like if I have a model that I quantize layers one, five, and 10, and another model where I only quantize layers one and It is possible that the model in which I quantized more information is going to perform better because the quantization errors have canceled out. And so what Joshua showed in his mathematical proof where he had like a verifier in, is that you can predict which layers are going to have quantization errors that will cancel out with each other, and you choose to quantize those layers. And so the result of doing this mathematical quantization is you end up with a model that's 20% more quantized than another provider, so you get 20% more throughput of it because there's more layers than running an NVFP4, and your quality is better than that other quant because the layers that you chose to quantize have their errors cancel out, like one layer skewed to the right one layer skewed to the left, one layer skewed to the right. Your final logits distribution is more similar to the original distribution of the model, so you have better fidelity. And so the way we proved this was with KL divergence. So instead of just scoring on the benchmarks, we scored the KL divergence between the logit distribution of the quantized model and the logit distribution of the original full precision model, and we showed that with this technique we get. If your probability distribution on the logits which token it wants to select is more of the same as the original model, you're probably gonna end up staying true to the original model. So yeah, so it seems like previously before this, it seemed like the industry was, well, the more you quantize, the worse it's gonna be, ‘cause the more loss you introduce. That's not exactly, not necessarily true. So yeah, doesn't improve it, but can cancel out.Philip [00:31:57]: I think it might be this, but reminds me a good bit about pruning where you can prune off certain layers.Philip [00:32:03]: But very interesting. Didn't know this was a whole paper you guys put out.Ali [00:32:06]: It's. Fun fact, it was originally 72 pages, this paper, and then we decidedPhilip [00:32:11]: WowAli [00:32:11]: We can't tell. We couldn't release it. So it's now 45.Swyx [00:32:15]: Still 39 pages, so very substantive. We talked about evals and all these things and, like what's possible in terms of speedup? Like it's like probably like the numberInference Speedups and BenchmarkingSwyx [00:32:25]: Thing that people do wanna care about, and it's something that you wrote about in your post. Like official API is 70 tokens per second, and you push it up to 90. Is that like a normal thing?Philip [00:32:36]: So what's cool about working in inference, the reason that I think inference is going to be a useful place to do engineering for a long time, is that if you look at highly optimized domains like, say, finance, if you're in finance, you measure how much better you got in basis points. It's like, “Oh, I got five basis points better, like twentieth of 1% better,” that's huge news because everything is so optimized. When we publish optimizations, it's 20%, it's 100% it's 200%. So there's still probably like a lot further to go, honestly. Like you'll, you'll know that inference is pretty much solved when researchers start publishing about how they got 1% faster at something.Swyx [00:33:19]: Which by the way, because I am from the finance background, in the ‘70s, that was the margin at the time. When you did quantitative finance research, you would findAli [00:33:27]: And like 20%, tens of percent.Swyx [00:33:29]: That's. Yes.Philip [00:33:29]: Yeah.Swyx [00:33:30]: And now it'Philip [00:33:31]: Tiny fractionsSwyx [00:33:32]: For those people interested, look up Andrew Lo's paper. He had a really interesting illustration of quant, stat arb, distribution, narrowing down from like those kinds of 20% differences in the ‘70s, down to nothing today, which is very cool.Philip [00:33:48]: Exactly, and we're at the beginning of the same type of thing. Now benchmarking is hard. I think anyone will tell you that, and benchmarking provider speeds is hard because there's so many variables that go into it. What hardware are you using? How much load do you have on the system? What's the exact nature of the prompts and input and output sequence lengths? All that stuff. But overall, when you start stacking these improvements, you're looking at multiples. You can look at it. The most common form, of course, is TPS, tokens per second, which is bad naming by us in the industry, ‘cause there's two tokens per second. There's tokens per second, the throughput number, and the latency number.Ali [00:34:31]: TTMT, yeah.Philip [00:34:32]: Like total tokens per second out of the, out of the GPU as a throughput number. Most people only care about tokens per second as the latency number, which we should call ITL, intertoken latency, but we don't.Philip [00:34:44]: Anyway, so you can imagine a standard API without many optimizations for a 1 trillion parameter model operating somewhere in the 30 to 50 tokens per second range for reasonable traffic profile. And we generally see the goal of, pushing to 10X that. But, not necessarily day zero, but by stacking enough optimizations, if you have, say like four optimizations, each of which doubles performance. Or sorry, three optimizations, each of which doubles performance, then you stack that up, that's an 8X gain. That's the order of magnitude that we're working with in this space. We're trying to make things substantially faster, not just go from like 70 to 90.Swyx [00:35:38]: Are you saying you've. You have done that?Philip [00:35:40]: So let's say you have as a reasonable baseline, 30 or 40 tokens per second. You can achieve 10X that. So like on GLM-5.2, if you run it unquantized, perhaps on H100s even, and you're just using an off-the-shelf inference engine with no particular optimizations, no speculator, nothing extra around like KV routing, no disaggregation, you're, you're probably, yeah, looking at that like 30 to 40. You think that's like a reasonable baseline?Swyx [00:36:12]: Right. Right.Philip [00:36:12]: To get to something like 10X, there's a lot of trade-offs that you're making. If we're running at more like a 300, 400 tokens per second range, you are using the best hardware possible. You have a optimized speculator. You have done all of your quantization work. You are Seeing a pretty high cache hit rate. You are running with a reasonably small batch size and a parallelism configuration that is tuned for latency versus throughput, but it is possible. So the spreads that you see if you, like, go on artificial analysis or you go on OpenRouter and you look at, the worst provider to the best provider, oftentimes can hit that range. 10X is of course very aggressive. It's oftentimes maybe more of a four to six times improvement. But that's the performance that makes us really excited, is when we can get these huge gains, not just go from 70 to 90 tokens.Stacking Optimizations: NVFP4, Speculation, and DisaggregationAli [00:37:19]: It's also, like, hardware dependent. Like, ifPhilip [00:37:20]: YeahAli [00:37:20]: If you have a thing where you're serving it on just, like, a node of H100s and then you throw, like, you shard the model across, like, four nodes of B200s. Like, you can definitely increase the speed with just throwing more hardware at it. Like, normalizing for the same exact hardware and the same number of GPUs.Philip [00:37:35]: Yeah. Then you're looking at, like, a two to 4X improvementAli [00:37:38]: Right. RightPhilip [00:37:38]: Depending on the inference optimizations. So yeah, it's. Some of it's, what's the call, and some of it's who's the driver.Vibhu [00:37:46]: If you break down the two to 4X, say the example is run GLM-5.2Ali [00:37:51]: YeahVibhu [00:37:51]: On B200sAli [00:37:53]: YeahVibhu [00:37:53]: Single node, right? What's, like, the cost trade-off for effort to get, like, the last bit of juice out versus what should people just think of, right?Ali [00:38:01]: Spectre quantization. Yeah.Vibhu [00:38:03]: Spectre quantization.Ali [00:38:04]: That's, that's, that's like 95%. LikeVibhu [00:38:06]: And how far does that get you? And how easy is that for the average person to do? So say right I wanna throw the weights of GLM-5.2 on a node of B200s, how easy is it to find speculative decoder- decoder model or already quantized model? How much work goes into it?Philip [00:38:23]: If you're doing it up front, it's quite a lot of work. If you're doing it today, there's going to be people who have published things that you can just, you can just grab some NVFP4 weights. You can grab a speculator. Yeah, if we're thinking about, like, what are the 2Xs we're stacking, going from, BF16 to NVFP4 is, it's not quite a 2X, right? It's like. I think it's about, like, 30 to 40%, from 16 to 8, and then another 30 to 40% multiplied from, 8 to 4. So that doesn't quite get you a 2X, but, like, roughly a 2X. Speculator, roughly a 2X. Disagg on top of that if you're able to get enough hardware and put enough traffic through it, another roughly a 2X. And then you add in some, double-digit percent increase from having just a better runtime with, the latest kernels and stuff behind it. And that's how it stacks up.Ali [00:39:21]: YeahPhilip [00:39:21]: So building each of those, like, building the, quantized weights is, for someone who really knows what they're doing, hours to days of work. Building the speculator, again, like, hours to days of work. And the, disagg setup, hours to days. Well okay, but like once you haveAli [00:39:39]: Once set up. Once set up. YeahPhilip [00:39:40]: Yeah, getting disagg working for the first time, I'm saying, of course, is very difficult.Philip [00:39:44]: The marginal implementationAli [00:39:48]: Like, if you're just grabbing, like if you are a person, like just a normal consumer who has access to, like, a node of B200s and you're wondering, “How can I just host it myself?” You don't need to quantize the model yourself. There's always gonna be, like, an open source quantized checkpoint. NVIDIA's gonna push one out if no one else does. You. Usually, the providers will have their own spec dec that they've trained as well. You don't need to train your own spec dec. You can just use that as well.Philip [00:40:09]: Yeah. Like, GLM-5.2 has its own MTP.Ali [00:40:13]: Right. Right.Vibhu [00:40:14]: What's multi token prediction?Philip [00:40:15]: Yes.Ali [00:40:16]: I'm justVibhu [00:40:16]: Can you explain that?Ali [00:40:16]: I'm just an expert.Ali [00:40:18]: I can do it for you in case I get it wrong?Vibhu [00:40:20]: No.Vibhu [00:40:21]: Yeah, you should correct if we're wrong, but their multi-token prediction can be used for self-speculative decoding.Ali [00:40:27]: I'm not sure. I'm not gonna correct that.Vibhu [00:40:28]: Okay. I'm semi-confident in thatAli [00:40:30]: Okay. YeahVibhu [00:40:30]: But someone can check. But it's useful to paint the story of, okay, not just the average person, but say a company wants to switch from serverless inference I wanna throw this up on. I wanna rent some GPUs, throw it up. These are the steps you take to do significantly faster than just put it behind vLLM.Ali [00:40:48]: Right.Vibhu [00:40:49]: I was waiting for a mention of Dynamo.Vibhu [00:40:51]: I feel like, that's supposed to be the baseline that you measure against.Dynamo, KV Routing, and Disaggregation ToolkitsPhilip [00:40:55]: I would think of Dynamo as less of a box system and more of a toolkit for building with. So when we talk about doing aware routing, when we talk about doing KV offloading, when we talk about doing, PD disaggregation, Dynamo fundamentally is. By the way, Dynamo is an open source library from NVIDIA.Ali [00:41:17]: We've done a pod with KylePhilip [00:41:18]: OkayAli [00:41:19]: Kyle Cranin.Philip [00:41:19]: Cool. So then your listeners know then that it supports all the different inference frameworks. And it is multi hardware, which is interesting.Ali [00:41:28]: But it's just a router, it's not like an optimizer layer.Philip [00:41:30]: Yeah. All it does, like, what Dynamo is good at, it is a library for moving information around your cluster, around your hardware. So if you have, KV cache on one place and you need it to be somewhere else, Dynamo coordinates NIXL for you to move that around.Philip [00:41:49]: That doesn't mean that, like, out of the box, you just say, “Pip install Dynamo,” and then you get, like, a massive performance speed up. It's more of a developer toolkit.Ali [00:42:01]: Yeah. I would have said it would. It comes with a set of defaults that you can then swap out.Philip [00:42:06]: It does. If the industry at large, I think, was, like, rolling out all of these deployments, standard, then I think it would be, like, a credible baseline. But, we've got to, we've got to benchmark against, like, what we're seeing in the wild.Speculative Decoding Methods: Medusa, EAGLE, n-Gram, and Spec-SpecVibhu [00:42:23]: I did wanna talk a little bit more about PD disagg, because that is probably, like, number three after quantized and speculative decoding. In your book though, I was just gonna pull out the book.Philip [00:42:31]: Yeah.Vibhu [00:42:32]: Like section 522 on Medusa, 523 on EAGLEPhilip [00:42:35]: YeahVibhu [00:42:36]: 524 on gram.Philip [00:42:37]: It's 55, would be disaggregationAli [00:42:42]: Yeah. Well, no, I just wanted to dwell a little bitPhilip [00:42:44]: YeahAli [00:42:44]: The other. Like, so what do you choose to include? What do you choose to not to include? Because there was all these other techniques.Philip [00:42:51]: Yeah.Ali [00:42:51]: Are these still relevant? Because I think they came out, like, a year and a half ago maybe.Vibhu [00:42:55]: Medusa is quite old.Philip [00:42:56]: Yeah, Medusa's old.Ali [00:42:58]: It was old.Vibhu [00:42:58]: But is it in the book as a good, here'sPhilip [00:43:01]: BaselineVibhu [00:43:01]: Baseline vanilla understand it?Philip [00:43:02]: Like you should know this.Vibhu [00:43:03]: Like I read the paper, I'm like, “ it makes so much sense.”Philip [00:43:05]: Yeah.Philip [00:43:05]: So with the book, I had a couple goals. One was to give people just a working vocabulary for the space as a whole, and the other was to give them some intuition about how each of these techniques works. As I mentioned in my AI Engineer talk, which is the first public addendum to this, the speculation space has moved much faster than everything else. So yeah, even at the time that I wrote the book Medusa, I very much included as a way for people to understand how the space evolved rather than what the most modern technique is. And now of course, there's DFlash, dSpark. There's, there's newer techniques even than EAGLE, although EAGLE is still very commonly used.Ali [00:43:51]: SpecSpecta.Philip [00:43:52]: Yes. Speculative decoding.Vibhu [00:43:54]: What canAli [00:43:56]: Oh, it's a paper by Tri Dao and it's like, it's doing speculative decodingVibhu [00:44:00]: HuhAli [00:44:01]: For the speculative decoder.Philip [00:44:02]: Oh, in spec- oh my God.Ali [00:44:02]: It's literally just an another. It's like, yeah, that's the most simple way to explain it, and it seems like he got trivial speed ups there. But it seems that the complexity with training, it's almost like in our mind at least, it's almost as complex as training GANs. Like it's like a very delicate balance and oftentimes you, it's just but yeah, it's literally speculative decoding on speculative decoding.Vibhu [00:44:21]: Speculative.Ali [00:44:22]: Yeah. We saw this paper.Vibhu [00:44:24]: It's interesting, right?Ali [00:44:24]: Yeah.Vibhu [00:44:24]: I wouldn't even expect it to be very particular to train, I wouldAli [00:44:29]: Right.Vibhu [00:44:29]: The naive part of me is like, okay, train speculative decoder.Ali [00:44:32]: But like, and it makes sense, like the whole idea of speculative decoding is you. It's like, it's like almost like the iPhone auto predict version but for a normal model, right? Like you're just, you're just, generating three tokens and you're like, okay, I'll do prefill on them. And so you save those three turns for your original model. Now your speculative decoder is doing three turns of auto regression, so why not just have an even smaller model?Ali [00:44:53]: The other question there is what are the size of speculators? So say forPhilip [00:44:58]: Right. It's like a billion parameters.Ali [00:45:01]: Like for MiniMax, it's. Yeah. It's like one layer. It's like one 60th of the original model usually.Philip [00:45:06]: Yeah. I think we should do a paper when we get back to the office.Philip [00:45:10]: SpeculativeAli [00:45:11]: SpeculativePhilip [00:45:11]: Decoding.Ali [00:45:13]: No, it's, it does seem like how, when do you stop? But then it also seems like if you're able to train spec-spec decode for instance, right? Like if you're able to have a small model that is accurately predicts what the intermediate speculator is gonna predict, that is able to predict what the original target model's gonna predict, then why not just use that smallest model directly, right?Vibhu [00:45:34]: Yeah. This isAli [00:45:35]: Like it seems likeVibhu [00:45:35]: Adjacent to the routing problem.Ali [00:45:36]: Right.Vibhu [00:45:36]: Yeah.Ali [00:45:36]: Right.Philip [00:45:37]: The thing with speculators is one of the practical constraints on using them is that you do have to run a small model on the same hardware that you're running the big model on. There is a orchestration and resource competition problem inherent in that, and that is one of the constraints on speculation in general, is that draft tokens cost resources to create and cost software complexity to manage. And so if you have like infinitely recursive speculators, you add in quite a bit of that complexity on the actual implementation within the inference engine as well, not just in the training process.Vibhu [00:46:17]: I was gonna say, I would wonder if you could do similar, like distillation and pruning of, it's the same thing, it's just a model. Can we not just distill a lot of the weights, quantize the speculator, out of my domain? The question that also comes up is, this is all for big server workloads, right? How much of this applies to, say I have this MacBook, I wanna run Gemma really efficiently. Similar problems, not the same?Local AI vs. Data Center InferencePhilip [00:46:45]: Pretty different. I talked to Selo, about this on his podcast a couple weeks ago. The difference between inference engineering for the data center and for production workloads versus inference engineering for local AI, is that we start with fundamentally like different constraints and different goals. With local AI, it's how do I fit this model onto my hardware and then make it less dumb? And with data center influence, it's how do I load this model and then make it less slow? And we care about less dumb, and they care about less slow. But the local AI inference engineering ecosystem, I think has a lot for us to learn from in the data center space. They are experts in various forms of quantization, including dynamic quantization that we just don't touch, in the pruning, in the distillation, in the, layer removal. There'Ali [00:47:42]: Layer removal matters less.Philip [00:47:43]: Yeah. There'Ali [00:47:44]: No one loves pruning really.Philip [00:47:45]: Yeah. Well, but the, but they doVibhu [00:47:46]: Which is surprising, right? But that's, that's a whole different thingPhilip [00:47:48]: Just to fit something on the laptop.Ali [00:47:50]: Right.Philip [00:47:50]: So yeah, it's a, it's an interesting, it's an interesting space. Not necessarily that like their techniques make sense for us to do in the data center, because we have different resources and different goals, but more that the process as well as the openness of that field is something to, admire.Ali [00:48:12]: Yeah. Like to your point, like, certain optimizations that would. Like for instance, Turbo Quantum Sharper, like it made such huge hype on that and we did like a whole deep dive on Twitter and like said, what is it? How does it work? Why is it good or not? And it took off and it was implemented on local devices because your memory bandwidth is so slow on like a MacBook, for instance. But try putting the same thing on like an NVIDIA GPU on a B200 Turbo quant would not be. Like, it would not be used. Like, NVIDIA - Like, NVIDIA made it clear that this is not a good optimization, and we've seen it firsthand where the overhead of doing dequantization, quantization of, in the kernel itself with turbo quant kernel, each end is much slower than the time that you save from doing the bandwidth. ‘Cause on the B200s, you have like 3.5 terabytes per second. You don't need decrease the storage that much. You don't need to do, FP4 KV cache. You don't need to use a requant. There's, there's, there's better optimizations to be made. But on Edge devices, it's extremely important, it's extremely useful. So, seems to be, like, different optimizations there, but then they're all uniquely combined with like all you wanna quantize the model, you wanna do speculative decoding, like certain common prefixes with bothPhilip [00:49:18]: Principles.Ali [00:49:19]: Yeah, exactly. Exactly. Exactly.Philip [00:49:20]: They also do a lot of work on, model parallelism, especially over, heterogeneous topology, where you have, some sparks and they are wired together with, Ethernet, DGX sparks.Ali [00:49:35]: Yeah, this is the Exo Labs guys.Philip [00:49:36]: Yeah. You have, a nu

Llevo 15 años escribiendo notas, artículos y tutoriales. El resultado: unos 5000 archivos markdown repartidos por mi disco duro. Y, como te puedes imaginar, encontrar algo ahí dentro es como buscar una aguja en un pajar. Por eso en este episodio me he puesto manos a la obra para montar un sistema RAG (Retrieval-Augmented Generation) 100% local, sin depender de APIs externas, sin enviar tus datos a la nube, y con herramientas que ya conoces: SQLite, Ollama y Python.Este es el primero de dos episodios sobre RAG. Aquí nos centramos en construir la base de conocimiento: un pipeline que escanea tus documentos, los trocea en fragmentos manejables, extrae los metadatos del frontmatter YAML, genera embeddings con el modelo bge-m3 de Ollama, y lo guarda todo en una base de datos SQLite con búsqueda FTS5. Todo esto, además, con detección incremental de cambios: la primera ejecución tarda lo que tenga que tardar, pero las siguientes son cuestión de segundos porque solo reprocesa lo que ha cambiado.El stack es sencillo pero potente. SQLite con FTS5 para búsqueda textual, Ollama con bge-m3 para los embeddings, y seis scripts Python que suman unas 1300 líneas. Nada de LangChain, nada de frameworks pesados. Código limpio, comentado y que entiendes de un vistazo. El chunking respeta las cabeceras markdown, usa tiktoken para contar tokens con precisión, y los embeddings se almacenan como BLOBs en la propia SQLite. En el próximo episodio (el 821) usaremos esta base de conocimiento para hacer búsqueda semántica con similitud de coseno, búsqueda híbrida combinando FTS5 con embeddings, y hasta un plugin para Neovim.Puntos clave del episodio:- El problema: 15 años de notas, 5000 archivos, cero capacidad de búsqueda- La solución: RAG local con SQLite + FTS5 + Ollama, todo en tu máquina- Chunking híbrido que respeta cabeceras markdown y usa tiktoken- Pipeline incremental con detección de cambios mediante MD5- Embeddings con bge-m3 (568M parámetros, 1024 dimensiones)- Búsqueda FTS5 con snippet(), colores ANSI y sintaxis avanzada- Errores comunes y cómo solucionarlosSi te gusta el contenido, ya sabes: dale a seguir, compártelo con quien creas que le puede interesar, y déjame un comentario si tienes dudas o sugerencias. La semana que viene, en el episodio 821, montamos la búsqueda semántica y el plugin para Neovim. No te lo pierdas.Capítulos del episodio:0:00 - Introducción: RAG y base de conocimiento local2:12 - El problema: 15 años de notas sin buscar4:58 - La solución: SQLite + FTS5 + Ollama, 100% local7:00 - Escaneo de archivos y extracción de front matter10:20 - Preparación del entorno: Ollama, uv y dependencias12:00 - Chunking: cómo trocear los documentos15:45 - Estructura de la base de datos SQLite17:46 - Pipeline incremental con detección de cambios19:22 - Demo en vivo: consultas y resultados22:44 - Errores comunes y cómo solucionarlos24:10 - Resumen y adelanto del episodio 82125:15 - Despedida y cierreMás información y enlaces en las notas del episodio

We've been raising funds with blockchain and crypto for a decade. But one team believed in a vision that is only now coming to reality. And you need to hear their story.Joris believed that Blockchains will create the ability for all assets to act as 'APIs' and unlock better capital, equity financing and growth... But to do so, he needed to wait for the rest of the world to catch up to his vision.Fast forward to 2026: the entire financial services industry is sold on Blockchain technology. Tokenisation of everything is ongoing. Even the regulators are pushing things forward. So what comes next?On this show we discuss:- An intro to Joris and Fairmint- The history of on-chain equity (the last 10 years)- Getting subpoena'd by Gary Gensler- Opportunities for on-chain compliance and data oracles- What more is needed to see wider adoption of onchain equityThis is one of the best shows we've done in Season 4, so don't miss it!

Why do less than a third of independent agents actually use a CRM, and what is that gap costing them? Jason Cass sits down with Mariah Gates, Founder of Accelerated Automation, to unpack agency operations, the shifting role of AMS platforms, and what it really takes to prepare an agency for AI. Key Topics: Why less than 27% of agents actually use a CRM, per a Vertafore study Growth through acquisition creates operational struggles even for large agencies Fixed owner pay leads to healthier cash flow and tech decisions The "AMS Demotion": AMS shifting from operating system to system of record AMS platforms need open APIs and two way sync to stay relevant Point solutions expose gaps that AMS systems fail to solve Why integration philosophy separates open platforms from closed ones like EZLynx Software companies must listen closer to users to avoid disconnected decisions AI adoption requires documented processes before layering on new tools Why agentic AI replaces unlicensed work while virtual employees remain essential Reach out to: Mariah Gates Jason Cass Visit Website: Accelerated Automation Agency Intelligence Produced by PodSquad.fm

Hoje o papo é sobre novos protocolos agênticos! Neste episódio, conversamos sobre como o Model Context Protocol (MCP) busca padronizar a integração entre agentes e ferramentas, além das diferenças e complementaridades entre MCPs, skills e APIs tradicionais. Vem ver quem participou desse papo: Paulo Silveira, o host que ainda acha automação uma bagunça Vinny Neves, cohost, dev e professor na Alura Mikaeri Ohana, Staff Developer Relations Engineer no Google Sulamita Dantas, Database Engineer e professora na FIAP Marco Antonio da Silva, Diretor de Engenharia do Conta Simples  Links:  Anthropic apresenta o MCP em 2024 MCP MCP Apps Registro oficial de servidores MCP Vinny: MCP tá onde o npm tava em 2014. e isso não é elogio Conta Simples + MCP MCP Tools Agent Gateway ADK: Agent Development Kit A2A: Agent-to-Agent UCP: Universal Commerce Protocol AP2: Agent Payments Protocol Google: Agent Identity Agent Skills Repositório oficial de Agent Skills da Anthropic  Plugins no Antigravity skills.sh, o npm de skills Criador de skills da Anthropic Roadmap do MCP Boas práticas de segurança para MCP Toda revolução tecnológica começa com quem antecipa o futuro e transforma ideias em soluções de alto impacto. Conheça os cursos da Alura + FIAP Skills & Go: Agentic Engineering, Building AI Products, e AI Data Strategy. Saiba mais sobre o Skills & Go. Vá para o Vale do Silício com Paulo Silveira, Marcell Almeida, Fabrício Carraro e Marcus Mendes na “Imersão IA Sob Controle e Alura no Vale do Silício“! Vagas limitadas, corra para reservar a sua. TechGuide.sh, um mapeamento das principais tecnologias demandadas pelo mercado para diferentes carreiras, com nossas sugestões e opiniões. #7DaysOfCode: Coloque em prática os seus conhecimentos de programação em desafios diários e gratuitos. Acesse https://7daysofcode.io/ Produção e conteúdo: Alura Cursos de Tecnologia – https://www.alura.com.br Edição e sonorização: Rede Gigahertz de Podcasts

Show DescriptionCodePen 2.0 is out now and we're talking about the launch, the idea of templating on the web, and how HTML could look very interesting in the future. Listen on WebsiteWatch on YouTubeLinks Yarn Announcing TypeScript 7.0 CodePen 2.0 announcement blog post SponsorsNotionWrite custom tools for Notion Agents that generate assets, query live data, and hit any API. Listen for incoming webhooks from any app, then run workflows with Notion Agents, pages, databases, and external APIs. All of this, on a hosted runtime. Workers are isolated sandboxes managed by Notion, so the code behind your syncs, tools, and workflows runs on our infra instead of your servers.

Don’t miss this massive SMB partner shift! Subscribe to our Newsletter: https://theultimatepartner.com/ebook-subscribe/ Check Out UPX: https://theultimatepartner.com/experience/ In this pivotal episode, we sit down with Jose Gomez Cueto, Microsoft’s SMB leader for the Americas, to uncover the monumental shifts happening within the partner ecosystem and the $20 billion cloud opportunity currently on the table. The discussion dives deep into Microsoft’s commitment to the CSP channel, the explosion of AI agents, and why shifting from traditional headcount growth to outcome-based results is critical for survival. From navigating the complexities of the marketplace to the urgency of becoming “Customer Zero” with AI tools, this conversation provides the roadmap every MSP needs to thrive in the new era of technology. https://youtu.be/QE-1w7GeyPM Key Takeaways Microsoft operates a $20 billion cloud revenue business in the Americas alone, with 80% driven by the channel. The Cloud Solution Provider (CSP) program is now Microsoft’s primary hero motion for the fourth region. The currency of SMB growth is shifting away from headcount and moving directly toward AI-driven outcomes. MSPs must transition from traditional IT outsourcing to strategic business process consulting to survive. Failing to proactively adopt and secure AI tools creates massive liability and shadow AI risks for organizations. IT providers are urged to become “Customer Zero” by deploying and testing Copilot and autonomous agents internally before selling them. If you're ready to lead through change, elevate your business, and achieve extraordinary outcomes through the power of partnership—this is your community. At Ultimate Partner® we want leaders like you to join us in the Ultimate Partner Experience – where transformation begins. Key Tags CSP, Agent 365, SMB cloud revenue, outcome-based selling, Copilot for business, Defender for business, shadow AI risks, AI agent deployment, Purview data security, Marketplace API integration, autonomous agents, Customer Zero, Microsoft Americas segment Transcript Jose Gomez Cueto AUDIO PODCAST [00:00:00] Jose Gomez Cueto: And, and you know, if I might say something that is confidential, avid Vince, uh, to be quite honest, please, please, uh, by definition, a marketplace is eliminating intermediaries. [00:00:11] Vince Menzione: You can feel it happening. [00:00:13] Vince Menzione: The ecosystem is shifting beneath us. The way Hyperscalers are partnering, how AI is remaking the channel and what it means to win in 2026. [00:00:23] Vince Menzione: Welcome to the Ultimate Partner Podcast. I’m Vince Menzi, own your host. And each week I sit down with leaders at the intersection of technology, partnerships and outcomes. The voices shaping how ecosystems actually work. We talk about what’s real, what’s changing, and what it takes to lead in this era where the partner channel isn’t just part of the strategy. [00:00:45] Vince Menzione: It is the strategy because being in the room changes everything. Let’s start. [00:00:55] Vince Menzione: I am absolutely thrilled for our, our next guest. Um, some of you heard me talk about this maybe earlier or in various pockets of conversation. Um, I believe both the SMB market is an, is an incredible opportunity. We’ve called it the Acre of Diamonds at Ultimate Partner at previous events. And then the MSP community, which I want to thank so many of you to for coming, coming on board now. [00:01:25] Vince Menzione: ’cause we’ve had some MSPs that have come to all our events. And doubled, tripled, quadruple the sizes of their business. From what they’ve learned in these rooms. And so we invited our next guest to come. Oh, Jose, come on up. Jose Gomez Cuerto is the leader of Microsoft’s SMB business for the Americas. Come on. [00:01:43] Vince Menzione: Come on over. Come on over. Sit down with me. And I was so thrilled to get this gentleman to come join us. His team is doing incredible work. I got to meet some of his team actually earlier this year. And we know each other for many years ago. [00:01:56] Jose Gomez Cueto: We do. [00:01:56] Vince Menzione: When I was at Microsoft, right? Yeah. So, so great to see you again. [00:01:59] Jose Gomez Cueto: It’s a pleasure to be here. Uh, thanks for the invitation. I’m thrilled to be here. And thank you all for making time, uh, or traveling here. Uh, this is the best time. To be in the industry. [00:02:10] Vince Menzione: It’s an incredible time. [00:02:11] Jose Gomez Cueto: Yeah, [00:02:11] Vince Menzione: it’s an incredible time. So sit down. Yeah, sit down. Let’s, yeah. So let’s talk about you and your organization. [00:02:17] Vince Menzione: Um, let’s talk, well, I, I, I wanna bring this up because it was like, the noise I heard in the room when I was, I went to Interven earlier this year. Yeah. Is, does Microsoft Care about this market? And, um, I was at Microsoft many years, we worked together when I was a, a gm. And, uh, it was run differently back in the day. [00:02:37] Vince Menzione: Yeah. And there’s been a lot of changes to what we call the SME and C business now. Mm-hmm. Uh, and the SMB business, which you run. So let’s talk a little bit about your organization, where you sit in the organization, and then I want to kind of dive in a little bit about what’s changed. ’cause a lot has changed for the better. [00:02:54] Jose Gomez Cueto: Yeah, it’s a great question and I think that that’s what a lot of people think about, uh, SMB and, and who’s SMB and, and who’s at Microsoft and who do I talk to. So, [00:03:02] Vince Menzione: yes. [00:03:02] Jose Gomez Cueto: Uh, even though I know a lot of, uh, friendly faces in the room, I think it’s a great, uh, starting point. Vince, so. Basically, uh, I am responsible for what we call the small and medium business, uh, segment. [00:03:15] Jose Gomez Cueto: Uh, we can also call it small and medium enterprises. Uh, I would say that it is not a monolith. Uh, we do have, uh, subsegmentation, I think that our friend Jay was talking about up to 20 subsegmentation. Uh, we think about it for simplifi simplification purposes on three. Uh, so we have, uh, the smaller organizations, the medium-sized organizations. [00:03:36] Jose Gomez Cueto: And then what we call top point manage, which is basically large enterprise that we simply don’t have an account management team, uh, assigned to. And we are the happy recipients of many of those, uh, every year. Uh, so I would say that, uh, a best definition would be also anything that is unmanaged and is primarily driven through the channel. [00:03:54] Jose Gomez Cueto: Uh, we run in the Americas approximately more than $20 billion of revenue, uh, on cloud. Uh, that’s [00:04:01] Vince Menzione: crazy. [00:04:01] Jose Gomez Cueto: So, and 80% of that is done. Through companies that are here, [00:04:05] Vince Menzione: $20 billion of business. [00:04:07] Jose Gomez Cueto: Yeah. So, um, the, the Americas region that I’m, uh, representing and under my responsibility includes basically three sales units, the United States, Canada, and Latin America. [00:04:18] Jose Gomez Cueto: Yes. Latin America is more fragmented because we have multi-country and multi, uh, subsidiary, uh, structure. Just to recap a little bit of what you asked me rewinding on what has happened in the last two or three years. Yeah. We brought basically, uh, probably something that you might remember from you were there. [00:04:36] Jose Gomez Cueto: I, yes. Uh, which is bringing the segment, uh, with the channel together. So, uh, I think that, um, uh. Earlier in the morning, uh, Steven was, uh, talking about it, what we call S-M-U-N-C, which is, uh, this segment with the channel. And the main reason is to drive, uh, that synergy, uh, and making, uh, a very bold statement that many of you might remember in the last two years, uh, Judson and Ralph, uh, heter or, or new, uh, president for this, uh, fourth region. [00:05:05] Jose Gomez Cueto: Uh, ’cause we call it fourth region. Yeah. ’cause the other one is our enterprises. [00:05:08] Vince Menzione: Yeah. So Asia, Americas Exactly. And, and EMEA. Then you’re the fourth region. [00:05:13] Jose Gomez Cueto: We’re the fourth region. So, so, um, making CSP, uh, our hero motion, and that is fantastic news. I, I started, uh, part of my journey, uh, in, in the channel, uh, way earlier in distribution in the year 2000. [00:05:30] Vince Menzione: Yep. [00:05:30] Jose Gomez Cueto: Uh, and fast forward, I would say 2011, we were launching the first commercial SaaS offering, which was Office 365. Um, I had the privilege to be, uh, leading the launch globally for that. Uh, but then we, the first thing we did was build a channel, and that was called syndication. And basically the precursor of that, uh, became CSP, basically putting, uh, the partner or customer in the middle, the partner around it for the, not only the, the opportunity, but also the responsibility to serve the customer. [00:06:04] Jose Gomez Cueto: 360 from, uh, presales all the way to, uh, uh. Upsell cross sell, and in between deployment, uh, things, uh, around, um, servicing, bundling offers, uh, troubleshooting and support, et cetera. So, uh, back to your question was, this is a very important thing because we’re basically, uh, making our channel the scale and, and the vision that we have is, is that we are gonna be continuing to scale through the channel. [00:06:33] Jose Gomez Cueto: So, um, one last thing I say, uh, in terms of the organization that I think is important for everyone to understand, and I’m gonna go a little bit into more org structure, is that we, we have these three sales units that are geographic. But, uh, what we’ve done this year is to have, uh, more depth on the solution area. [00:06:50] Jose Gomez Cueto: So you might remember that, uh, we’ve simplified them, uh, same as we used to have 8, 13, 13 areas. Now we have only three Oh yeah, same, same, uh, in the solution area. So we have, uh, the AI business solutions. Uh, cloud and AI platforms. And then, uh, security and my team basically mirrors that structure. And we have, uh, team members, uh, primarily, um, our partner, solutions specialists that are, their job is to work with companies like you. [00:07:17] Jose Gomez Cueto: Uh, some few we do, uh, direct in others. What we do is work with, uh, our top distributors, and I think we have, uh, in the room many of those. I think Google is gonna follow up, uh, for PAX eight, and that’s, uh, how we’re going to market now. [00:07:31] Vince Menzione: So just a little bit of context too for me. ’cause I, I, I had heard this at another event. [00:07:36] Vince Menzione: Yeah. And I just wanted to share this. Um, when I was at Microsoft, we, we did not put the right emphasis and energy and resources in the s and b market when I was there, or, or it was fragmented. Every group did it differently. You remember those days too, right? Well, with public sector, we didn’t necessarily have a team focused, and every business did it a little bit differently. [00:08:00] Vince Menzione: Mm-hmm. And I think one of the contexts you, you mentioned Ralph and being in Ralph’s organization. Yeah. Pulling that all together and creating the fourth region created a lot of focus that didn’t exist. And consistency in terms of execution. I think that’s what you’re talking about here. Right? And then also the fact that like, we didn’t, I don’t think we had a sep, an SMB leader back in, back in the day. [00:08:22] Vince Menzione: Like we didn’t have somebody that we can go to to think about the MSP community the way we do today. Mm-hmm. Right. We were just, they were just almost like unmanaged entities out there. Yeah. Was that, would you, would you agree with that? [00:08:32] Jose Gomez Cueto: Yeah, I, we went through several iterations that, uh, you might argue, uh, were painful or not. [00:08:38] Jose Gomez Cueto: Uh, ultimately what we’re committed is to simplify the partner experience. And make that the same for the customers. But what we have is now one center of gravity, uh, a global SMB organization. We have three area leaders. And uh, and that helps us, uh, to be quite honest and in confidence. And you and I talked about it, Jose, this is a forum for, uh. [00:08:58] Jose Gomez Cueto: Speaking the truth. Uh, we, we, we have to fight the gravitational force of the managed space. The company has a big enterprise footprint, so, uh, many of us have become, uh, the chief agitators, uh, to fight the good fight, uh, for SMB. Uh, try to under unpack, uh, in every single conversation with senior Execut. [00:09:17] Jose Gomez Cueto: What is an MSP? And no, it’s not data consulting or one of the large, uh, global design. Uh, and then we explain what they do and then what is a two tier channel, how do distributors work? And, uh, and what about this and what about that? So I think that that has been, uh, a great, uh, progress and a lot of that can be reflected, uh, into how we’re, hopefully everyone in the room is seeing it in how we’re going to market. [00:09:40] Jose Gomez Cueto: I’ll give you two examples. [00:09:41] Guest: Yes. [00:09:42] Jose Gomez Cueto: Um, for, for quite some time. We, we have very limited, uh. Product truth. That’s what the lingo that we use internally, uh, related to offer that were targeted to SMB. And I would say that, uh, business premium, uh, for M 365, uh, was the fact to offer. But now we’ve been able to in, uh, increase, uh, the not so not only commitment, uh, but also the investment that we’re doing as a company into launching offers. [00:10:08] Jose Gomez Cueto: So we have a co compiler for business that is. At a lower price point that has, uh, the same capabilities at the enterprise, uh, that we’re, uh, doing that we also have some security, uh, offers, uh, that are now unattached to business premium, which is our hero motion for sub 300 space. So you start to see, uh, an important trend and it’s great to have jobson at a CEO, uh, of the commercial business capacity because, uh, we’re making things happen. [00:10:34] Jose Gomez Cueto: So what I would say is that I love coming here to these forums. A lot of my team members are here. We’re here to learn. We’re the learner. All we, we, we don’t know much. We need to learn more. Uh, and, and just keeping us honest in terms of bringing that, uh, ethos of, of the customer that most of you are serving and, and, and things that we can improve to get better to deliver value. [00:10:58] Vince Menzione: Yeah. And the speed at which you’re moving has been pretty fast. It’s been very nimble. Like I, I, I’ve been watching this progression. It’s really like you, you’re really leaning in. I was actually hoping because I could ask you a bunch of questions. Yeah. But we have such a great audience and for the first time we really have opened it up to a lot of MSPs in the room. [00:11:18] Vince Menzione: Yeah. And I know you, you wanna get some interaction with some of these folks as well. I thought maybe we would open if you’re okay with this. Yeah, absolutely. I’d rather than I go off script a little bit. I’d rather open it up to some of the MSPs in the room. We’re sitting here eager to learn how and, and what Microsoft is going to do to help. [00:11:35] Vince Menzione: Because I think the opportunity, I personally think the opportunity is huge right [00:11:38] Jose Gomez Cueto: now. Yeah. Let’s do that and well, we get, uh, warmed up. I would say that. [00:11:43] Vince Menzione: So we need some mics. Yeah. [00:11:43] Jose Gomez Cueto: Uh, something that I’m, that I’m seeing, uh, Vince, and, and, and a question that many of you might have is why now? And, and why this an, an exciting, an exciting time. [00:11:54] Vince Menzione: Yes. [00:11:54] Jose Gomez Cueto: Um, and I would say that, uh. Right now we’re seeing, obviously Jay talked about it and, and the big transformation, but it’s a once in a generation or one in a lifetime. Yeah. Uh, shift of the entire platform. Uh, and, and a lot of the scenarios are even maybe scary, but what we see is huge opportunity. And from an SMB perspective, uh, the biggest thing that excites me is moving from, um, something that was. [00:12:23] Jose Gomez Cueto: More related to size, and now we’re moving to outcomes. So, so think about the future of SMBs, uh, with agents and things being measured on outcomes. And, and what this leads to is, uh, Jay talked about it as well, and sorry Jay, it’s such a good job that I keep quoting you. Um, we do that a lot. Uh. You got it. [00:12:49] Jose Gomez Cueto: So you talked about, uh, I noticed that Bill Gates when he said, you know, uh, uh, a pc, uh, in every desk and what we see is every human empowered with agents. Yeah. Especially in work. And what does that mean, that the currency changes being, because what you’re gonna be able to, to envision. Not in the, in the, in the near future, but now is an agentic explosion where then, uh, the currency is outcomes? [00:13:14] Jose Gomez Cueto: Yes. So if you think of an SMB growing, it’s not growing on, on, on full-time employees or headcount. It’s growing on the ability to do more through agents. So, so I think that’s an important thing and, and that’s something that we’re working very closely with our all, all our channel and the offerings that we’re launching to market as well. [00:13:32] Vince Menzione: I also think about the MSPs as being perfectly positioned because what you described, the new, the new model, the future customer and the outcomes is gonna require hands on the steering wheel at all times. [00:13:44] Jose Gomez Cueto: Yes. Yeah. So on that one, and still waiting for some, uh. Someone that is not shy to ask questions, but we’ll, we’ll keep going in the meantime. [00:13:52] Jose Gomez Cueto: Uh, I, I think that, uh, we are learning, all the [00:13:55] Vince Menzione: MSPs are lined up over here. I’m marching them all. [00:13:57] Jose Gomez Cueto: We, and, and I almost know by name all everyone in the first two rows. Yes. Uh, so, so, uh, I might pick on them. Uh, they’re too shy, but, but we’re learning together. Uh, Vince, uh, the important thing is, is the transformation, uh, and the opportunity, but also the risk of, uh, not acting. [00:14:17] Jose Gomez Cueto: Uh, what we were seeing, uh, for the first, uh, year or two was kicking tires, people testing, uh, ai. And now what we’ve seen is basically, uh, a full adoption. Uh, of the agentic technology, not even adoption of the tools, but embracing the technology. So I, I want to give you, uh, two specific, uh, examples or data points we have, uh, just in the Americas, more than almost 9 million, uh, people using copilot chat. [00:14:49] Vince Menzione: Wow, that’s amazing. [00:14:50] Jose Gomez Cueto: So imagine, uh, the, the potential that is there for people that are actively using the tool. Yeah. Uh, to en enable new scenarios of doing things. Uh, another example, and I think I have, uh, someone in my team here, is Amber in the room. Amber Kinney? No, she left. Okay. So Amber runs, uh, cloud and ai, uh, uh, or Azure platform. [00:15:12] Jose Gomez Cueto: Uh, her team has deployed, uh, more than, uh, 11 agents internally for our partner solution specialist, uh, from. Simple agents that will, uh, tell is if a specific deal is eligible for a pre-sales or post-sales program. And comparing all the complexity of our programs, oh my [00:15:29] Vince Menzione: goodness. [00:15:30] Jose Gomez Cueto: All the way to, to, to managing a pipe more effectively of opportunities. [00:15:34] Jose Gomez Cueto: So what we’re seeing is real. This is not something that people are just kicking the tires. It’s like this is the opportunity. So back to, to the point of m ms. P uh, is, is about learning together on how to transition. To, uh, a model that is gonna be based on outcomes. And, and we were discussing, uh, I was with some of our distributors, uh, many of them in the last two months in, in a specific partner advisory, uh, councils and, and some people were just sharing their experiences. [00:16:04] Jose Gomez Cueto: Oh, I decided to charge X amount for an agent. And how do you come up with that number? I don’t know. We’re just testing. Okay. And what about their current revenue? Uh, and, but what about the tokens? What if, uh, the agents start to consume and they’re gonna do the metering? So, so I think that we’re learning together in this space. [00:16:22] Jose Gomez Cueto: Um, but what it is important is just to think about the important, the, the, the critical role that the MSPs are gonna have in leading. And the biggest challenge that we’re seeing and, and we see it over and over and over is, uh, the part about scaling. [00:16:38] Vince Menzione: Yes. [00:16:39] Jose Gomez Cueto: The skilling is not, uh, about learning how to use the copilot tool or to do, uh, some, uh, you know, tuning and that, because thankfully our, at least our, our technology as a platform, uh, pretty much carries the same, uh, security, uh, and compliance configurations that you have in your Microsoft 365 tenant. [00:17:00] Jose Gomez Cueto: But it is more the, the, the skilling about understanding how to do. Customer outcome conversation. What is your AI strategy? What [00:17:08] Vince Menzione: that’s scaling? Yes. [00:17:09] Jose Gomez Cueto: What really matters? Not [00:17:10] Vince Menzione: the technical skill. It’s, it’s really the approach that they’re taking. [00:17:14] Jose Gomez Cueto: Yeah. [00:17:14] Vince Menzione: With the organization. I, it seems that MSPs for many years were down in the weeds. [00:17:20] Jose Gomez Cueto: Yeah. [00:17:20] Vince Menzione: They were turning the, the wrench, so to speak, in the organization, and yet now it seems like this. Kevin Piker, your old boss used to use this term. The, the CIO. The CEO is the new CIO. In other words, you need to be selling upstream. You need, you need to be having the conversations in the organization that are strategic [00:17:40] Jose Gomez Cueto: Yeah. [00:17:40] Vince Menzione: To that organization. [00:17:41] Jose Gomez Cueto: So, two, two twofold on, on that, uh, point, which is very important. One is, uh, not our, a lot of our MSPs are equipped right now. [00:17:49] Vince Menzione: Yeah. [00:17:49] Jose Gomez Cueto: To have a, a conversation about business strategy. Because traditionally has been more outsource it. [00:17:56] Vince Menzione: Yes. [00:17:56] Jose Gomez Cueto: Uh, we started with, you know, managing the networks, then adding services, support tickets, et cetera. [00:18:03] Jose Gomez Cueto: So being able to have that conversation is important. Uh, we, we see through a lot of our tooling that, uh, the shadow AI is everywhere. And what I always tell in any MSP conversation that I have is risk security. You’re on the hook if something happens. That’s right. So if you’re not acting. Uh, then it is a liability. [00:18:22] Vince Menzione: You’re letting things take off in your own organization. Yeah. People are using [00:18:25] Jose Gomez Cueto: philanthropic on their own. The company can go, uh, bankrupt or get sued or get, uh, if they’re in a regulated industry, they can be taken out, et cetera. So, so that’s an important point, uh, related to, to that transformation. Uh, and, and the other part of the skilling that you mentioned that is super important is being in the weeds. [00:18:45] Jose Gomez Cueto: That is where the innovation is happening. Yeah. The later research that we have is being in the front line because it’s all about, uh, reinventing those processes. So I think that it’s a, it’s a good combination that if we have the MSPs, um, and we’re working, uh, not only internally but with our distributors to develop the right skilling around those other type of, uh, consulting skills. [00:19:07] Jose Gomez Cueto: Uh, data skills, uh, business process, uh, redesign and flows. Uh, that is where, where we see the big opportunity. [00:19:14] Vince Menzione: So it’s balancing out the technical skills with the business process skills, the consulting skills. Yeah, exactly. I think we have a question over here. Yeah. [00:19:22] Guest: Good afternoon, Vince. Good. Sorry. Thanks for the great content. [00:19:26] Guest: The question is around small medium businesses and the cost around cybersecurity. So. Basically, as new tools are coming up that are AI based, such as co-pilot for security, defender for AI, are also consumption based, is there a risk that SMEs will be left out under that cybersecurity poverty line? [00:19:52] Jose Gomez Cueto: I don’t think, uh, it is, uh, a risk to being left out, uh, in the country that the, the SMBs, I would say are more help is needed. And, and the way we think about it from a perspective of, of ai and specifically I’m want to talk about agents, uh, it was mentioned by Steven in, uh, in the morning, and I’m gonna talk a little bit high level and then I’m gonna try to bring it down to, to more tangible is this concept of intelligent and trust. [00:20:19] Jose Gomez Cueto: So on the intelligence, what, what we’re, what we’re trying to say here is that your AI is not just generic stuff that you just prompt and you get like anything that is on the web, but there’s contextual. Data, and, and that’s what we do, uh, with what you might be familiar with, which is the iq. Uh, so we have, um, iq, uh, also in Foundry and on our different data products. [00:20:40] Jose Gomez Cueto: So basically bringing the context of your work, of your contacts, of the people you interact, uh, of the meetings of, of the emails, of the SharePoint files, but also important connectors that are in line of business applications that you can bring to copilot. And then. That intelligence, uh, is relevant and that that basically increases innovation. [00:21:01] Jose Gomez Cueto: And the part about trust, uh, uh, not exactly in cybersecurity, but, but related is basically, uh, agent 365. Uh, can I see, show of hands, who’s aware of Agent 365? Maybe like [00:21:14] Vince Menzione: in the front two rows, [00:21:15] Jose Gomez Cueto: 20%? Yeah. So, um, that is basically, uh, an, an amazing opportunity for our MSP channel because it gives you opportunity to. [00:21:25] Jose Gomez Cueto: Basically observe, uh, govern and apply security to the, the agent activity that is happening. So we think in the context of ai, I think that that’s a, a, a super important, uh, aspect to mitigate any risk of, of what can happen if there’s not, uh, the right, uh, posture. Uh, and then, uh, on, on, on the other part of security, I would say that something, I mentioned something about offers. [00:21:51] Jose Gomez Cueto: We brought the capabilities of the enterprise, uh, SKUs and solutions into these add-ons to N 365. So I would say that with, uh, defender for business, uh, plan two, and sorry to go into the SKU language, uh, it, it is important to, to understand that you have those advanced capabilities. And then another one that we’re pushing, uh, hard and, and is had great receptionist, um, uh, purview, uh, and purview. [00:22:15] Jose Gomez Cueto: What allows you is just to really do everything related to data. Data security policies of what data should be prompted by the model, what information to stay or, or, or not stay. Uh, and I think that’s, that’s also a good opportunity that we’re seeing to bring those, uh, advanced capabilities into the SMBs. [00:22:33] Jose Gomez Cueto: The challenge that we have is how do we get them faster, uh, to everyone, especially when there’s, uh, you know, competing, uh, so solutions around it. [00:22:44] Vince Menzione: We have one more question, and I think we’re probably gonna have to break after that. I know we’re over time already and you’ve got a busy rest of your day. I got, well, we got one back there and we’ve got a mic up here, so, so we have two questions. [00:22:57] Vince Menzione: Yeah. We’ll do Tim first and then we’ll get the [00:22:59] Vince Menzione: mic up. I’ll go for the first 30 minutes and we’ll go from there. Yes. Long time listener. Great to see you again. Jose. Um, business premium, we did E seven. We talked about getting a voice from the MSP space. To build out a business premium, like additional offering. [00:23:13] Vince Menzione: Is there any context to that you have any vision in your crystal ball for October? [00:23:17] Jose Gomez Cueto: Uh, I cannot say or, or deny. Uh, but yeah, I think that what, what you I love it in, in all seriousness team. Uh, thanks for the question. Uh, I think that what you should expect is, uh, I call it product truth, uh, more, uh, SMB built purpose built for solutions. [00:23:35] Jose Gomez Cueto: So an equivalent of, of any seven as well. Yeah. [00:23:40] Vince Menzione: You still have, we have another question in the back? Yeah. Yeah. Okay. [00:23:43] Guest: Yeah. Uh, Jeremy here with Integral, um, there’s this kind of idea going around that while CSP has been very successful for many of us as MSPs and, and since the beginning, it’s been a great program that was focused on s and b and it’s come up now. [00:23:57] Guest: There’s this kind of shift saying, and CSPs and you think about being marketplace companies where CSP is, the plumbing and marketplace is, is the lead. If that is true, or maybe you comment on that, that idea. How does marketplace strategy playing into kind of, I guess I’m plugging serials piece now from behind, but how does marketplace strategy then play into the s and b market if CSPs are focused on that marketplace mechanism? [00:24:21] Guest: Where CSPs now are and the, and the modern work and all the things that we’ve been doing really well for a long time become, maybe plumbing is too far down the stack, but really marketplace being a focus, is that a strategy piece that we should be thinking about for CS p strategy overall? [00:24:36] Jose Gomez Cueto: Yeah, it’s a great question and I’ll try to keep it brief. [00:24:39] Jose Gomez Cueto: Uh, I think you need my v The vision that we have is we’re doing both. Uh, we’re empowering, uh, and customers to find what they need. Uh, in the marketplace. Uh, zero talked about also the opportunity for resellers to get enrolled and start to add services and other things. There’s also, another part of the is, is multifacet, uh, to work with ISVs to make it easier and recruit them to bring the right offers. [00:25:03] Jose Gomez Cueto: For SMBI would say that the feedback that we need is to make sure that the right SA ISVs are the ones that serving SMV are represented. Then from another front, I would accept that yes, we have some, uh, plumbing work to do because right now, uh, some part of the billing is not really that nimble for a two tier model if you’re working through a distributor. [00:25:23] Jose Gomez Cueto: So we made some great progress. Uh, we, we, uh, have, uh, announced something and Ignite, if you missed it, I think we might talk about it, uh, soon. Uh, but we have that, that connection via APIs with, uh, the four largest, uh, global distributors. So we’re making progress towards something that will be seamless. Uh, but I think that the biggest opportunity that we have is, is to crack the code, uh, for marketplace. [00:25:46] Jose Gomez Cueto: And, and, you know, if I might say something that is confidential, avid Vince, uh, to be quite honest, please, uh, by definition a marketplace is eliminating intermediaries. So that’s the dilemma. How do you bring the channel in between to help you expand, [00:26:02] Vince Menzione: right? [00:26:03] Jose Gomez Cueto: That that is really the, the, the, the, the holy grail, if I may use those words. [00:26:07] Jose Gomez Cueto: Uh, but that’s something that, that we’re working towards. And I think, uh, we have a great opportunity ahead and, and you should expect, uh, more announcements as we head into the summer events on how we’re gonna make that more seamless. [00:26:19] Vince Menzione: And REO really lit up the channel Yeah. In, in a big way. ’cause that a hundred percent, that was a blocker before. [00:26:24] Jose Gomez Cueto: Yeah. [00:26:24] Vince Menzione: Yeah. But CSP is also an incredible opportunity if it, you know, I know, I know there’s other sessions and conversations around it. And it does feel, and I’ve heard this before, like I wanna buy from my MSP because they’re the ones I trust. [00:26:37] Jose Gomez Cueto: Yes. [00:26:37] Vince Menzione: But yet I go, I have to go around the system in order to transact my Microsoft licenses. [00:26:43] Vince Menzione: Right. Yeah. And that’s, [00:26:45] Jose Gomez Cueto: I think the scenario getting the gentleman was mentioning is related to marketplace. But yeah. Vince, uh, uh, I just wanted to perhaps close, uh, please. Because I think we’re outta time, right? Yeah, we [00:26:54] Vince Menzione: are. [00:26:54] Jose Gomez Cueto: Yeah. Uh, just in terms of what to expect, uh, we are continuing to be, uh, partner centric. [00:27:01] Jose Gomez Cueto: You should expect as we go into the next fiscal year, uh, more refinement into the customer subsegmentation, we have this concept of above 300 and below 300, uh, working even closer with our distributors to help us scale and amplify the efforts that we do around recruitment, scaling, go to market, uh, co-sell, et cetera. [00:27:22] Jose Gomez Cueto: Uh, and then, uh, obviously expect, uh, we, we, a call to action that I have for everyone is become customer zero. Vince, I’m gonna put you on the spot here. How many agents did you use today? [00:27:37] Vince Menzione: None. [00:27:37] Jose Gomez Cueto: Okay. [00:27:38] Vince Menzione: I, I’ve been in the room leading the room today, [00:27:41] Jose Gomez Cueto: even with more reason. [00:27:42] Vince Menzione: No, I, in, I need to do [00:27:43] Jose Gomez Cueto: more. Put your autonomous agents. [00:27:44] Vince Menzione: I do. [00:27:45] Jose Gomez Cueto: I’m not kidding you and I didn’t, I need to be [00:27:47] Vince Menzione: more of [00:27:47] Jose Gomez Cueto: a frontier for myself. The answer I get usually is like one hand raiser, by the way. Uh, but, but, uh, jokes aside, uh, I think. Becoming customer zero is critical. We cannot be deploying and selling what we’re not using. Uh, we have, uh, great tooling for low-code scenarios, uh, in, in, in, now, I don’t wanna say like in a few months now we have no one, uh, people that have zero knowledge and coding already developing and deploying agents into a secure environment. [00:28:19] Jose Gomez Cueto: It is happening. [00:28:20] Vince Menzione: Yeah. [00:28:20] Jose Gomez Cueto: So, uh, then, uh. Copilot. It is not a competitor charge, GVP or cloud. It is a platform we have both included. [00:28:29] Vince Menzione: Yes. [00:28:29] Jose Gomez Cueto: Do we have multimodal, we have iq. That is everything, uh, closed in terms of, uh, your intelligence. It is secure by default. Uh, and then allowing you to, to do, um, agents and then agents 365 to manage it. [00:28:41] Jose Gomez Cueto: So basically those three stages, customer zero. Uh, copilot agents and Agents 365 as your tool to, to manage them [00:28:50] Vince Menzione: and don’t go rogue and start doing your own things with anthropic and setting up your own instances because you’re gonna compromise your, your instance in your environment. [00:28:59] Jose Gomez Cueto: Well, actually, uh, if you do it in the copilot interface [00:29:02] Vince Menzione: Oh, well, I’m saying do it. [00:29:03] Vince Menzione: Yeah. I’m, I’m at RO going off, off, off, uh, [00:29:06] Jose Gomez Cueto: off. Yeah. Yeah, [00:29:07] Vince Menzione: yeah. Great. Well, thank you, sir. Appreciate you. Thank you. Thanks for listening to the Ultimate Partner Podcast. If today’s conversation resonated, share it with a partner leader in your network. Subscribe where you listen, and head over to the Ultimate partner.com for show notes related content and the resources for this episode. [00:29:29] Vince Menzione: And if you haven’t already, now’s the time to register for the Ultimate Partner Live event in Reston, Virginia, October 26th through October 28th. Until next time. Keep showing up in the rooms that matter because being in the room changes everything.

Alexandra Lato, VP of Product at Stack Overflow, joins Scott to talk about how Stack Overflow is evolving for the age of AI agents. They explore the new "Stack Overflow for Agents" feature, where a single prompt bootstraps an agent with all the knowledge and APIs of Stack Overflow, and what it means for how developers, agents, and communities share and verify knowledge together. https://agents.stackoverflow.com/

In recent months, the open vs closed, and US vs China discussions on model ownership and sovereign/local AI have heated up to a fever pitch. So it is very very good news that Poolside AI are finally emerging with new models, like Laguna S 2.1, that are beating Thinking Machines' recent release nearly 10 times their size.Poolside's recent tech report got a lot of praise due to their level of detail, and Vibhu first covered Laguna's recent technical report on our paper club:From spending $12 million building language models for code before the world cared to creating a Model Factory that can take a model from pre-training to release in eight weeks, Eiso Kant has spent more than a decade betting that code is the path to AGI. In this episode, the Poolside co-founder joins swyx and Vibhu to explain why ChatGPT felt like vindication, why Poolside embraced open weights and open research, and why he would rather live in a world with 100 foundation model companies than five even if Poolside were one of the five.We go deep on Poolside's Model Factory: the engineering systems behind 10,000–20,000 experiments per month, streaming data directly into training, reproducible experimentation, low-precision compute, and agents that increasingly write code, launch jobs, evaluate results, and modify the pipelines used to train future models. Eiso also unpacks their recent launch Laguna S, why persistence, verification, and backtracking may matter more than raw intelligence, how much capability remains inside smaller models, why reinforcement learning will move earlier into pre-training, and why next-token prediction is still extracting too little from the web.We also discuss model-harness co-design, Poolside's path from coding agents to AGI, why Eiso thinks MCP and traditional tool calls are “stupid,” the real economics behind frontier-model training, Poolside's $500 million raise, open-source AI, regulation, NVIDIA and TSMC's influence, engineering productivity in the agent era, high-agency teams, and hiring at Poolside.We discuss:* How Andrej Karpathy's RNN work inspired Eiso to start building language models for code in 2015* Why Eiso spent four years and $12 million pursuing an idea before the market cared* Why ChatGPT felt like vindication and brought Poolside back to open source* Why Eiso would prefer 100 foundation model companies over an oligopoly of five* The difference between releasing open weights and publishing genuinely open research* Why Poolside deliberately built a global research organization outside the Bay Area talent war* Why model building is ultimately 90% engineering* The Model Factory: Poolside's end-to-end system for rapidly training and improving models* How fewer than 70 researchers run roughly 10,000–20,000 experiments each month* How Poolside moved from six-month model cycles to five- and eight-week launches* Why streaming data directly into training unlocked faster experimentation* How immutable data, versioned code, and reproducibility enable rigorous model research* Why Eiso wants capable researchers to leave their labs and become Poolside's competitors* Why 95% of model building can be reduced to better data or compute efficiency* Laguna S and why persistence, verification, and backtracking can outperform raw intelligence* Why smaller models may handle far more knowledge work than previously expected* Why reinforcement learning will move earlier into pre-training* Why next-token prediction is still failing to extract enough knowledge from the web* Why distillation and environments have become the AI industry's favorite “drugs”* Why mid-training is really an early form of curriculum design* Low-precision training, networking bottlenecks, and the next gains in compute efficiency* Laguna S: 118 billion total parameters, 8 billion active, and eight weeks from training to launch* Why model builders can often evaluate a new checkpoint within its first 30 minutes* Model versus harness: where agent capabilities actually come from* Why Poolside sees coding and long-horizon software tasks as a path to AGI* Why Eiso thinks MCP and traditional tool calls are “stupid”* Why future agents will write scripts instead of choosing from dozens of predefined tools* The case for minimal harnesses, containers, and model freedom* Why Poolside is prioritizing vision but does not expect to work on audio soon* Why language may be the most compute-efficient modality for encoding knowledge and reasoning* The real cost of model development and why the final training run is anticlimactic* The story behind the Poolside name and why it represents refusing to lower ambitions* How Poolside raised $500 million while investors still questioned whether AGI was real* Why intelligence could become the world's most demanded and commoditized resource* When open models may become too capable to release without restrictions* Why unilateral AI safety does not work in a globally competitive environment* How regulation could accidentally lock in an oligopoly of two or three AI companies* NVIDIA, TSMC, and the hardware systems underpinning foundation-model progress* Why reinforcement-learning wall-clock time is one of Poolside's biggest bottlenecks* Why Poolside trains models from scratch instead of simply distilling larger models* How AI changes the way companies should measure engineering productivity* Why agency may become the most important quality for employees in the AI era* How leaders align high-agency people through shared goals and clear constraints* Hiring across research, post-training, pre-training, architecture, evals, and engineering at PoolsideEiso KantLinkedIn: https://www.linkedin.com/in/eisokantX: https://x.com/eisokantPoolside: https://poolside.aiTimestamps00:00:00 Introduction00:00:54 Karpathy, RNNs, and Building Code Models Before Transformers00:02:26 The $12M Failure and ChatGPT Vindication00:03:39 Open Source and the Case for 100 Foundation Model Companies00:09:22 Open Weights, Open Research, and Poolside's Global Team00:16:04 The Model Factory: Why Model Building Is 90% Engineering00:20:19 Agents, Automated Experiments, and Early Signs of RSI00:24:04 Streaming Data, Reproducibility, and Scientific Rigor00:30:35 Creating More Foundation Model Companies00:36:07 Laguna S: Persistence vs. Raw Intelligence00:43:01 Reinventing Pre-Training, RL, and Curriculum Design00:52:33 Low-Precision Training and Squeezing More From Smaller Models00:58:37 Model Harnesses, Coding Agents, and the Path to AGI01:09:26 Why MCP and Traditional Tool Calls Are “Stupid”01:13:04 Vision, Multimodality, and Why Language Still Matters01:18:15 Scaling Models and the Real Economics of Training01:20:40 Why Poolside Is Called Poolside and Raising $500M01:27:37 Open Models, AI Safety, and the Risk of an Oligopoly01:33:53 NVIDIA, TSMC, and the Reinforcement-Learning Bottleneck01:41:52 Smaller Models, Distillation, Engineering Productivity, and HiringTranscriptIntroduction: Eiso Kant, Poolside, and Open ModelsSwyx [00:00:00]: All right, we're here in the studio with Eiso Kant from Poolside, together with Vibhu. Welcome.Eiso Kant [00:00:08]: Thanks. Thanks for having me, guys. Good to be here.Swyx [00:00:10]: Yeah, fresh on the plane. You texted me, you were like, “Hey, I'm on my way to SF.” I was like, “You're on a plane right now, right?” Like, hey.Eiso Kant [00:00:16]: I know. After I texted you, I realized that probably coming in with major jet lag was gonna offer some fun experiences today, but let's do it.Swyx [00:00:23]: I mean, I think the thing I would tell guests is that they don't have to prepare that much because if you're truly working on this every single day, then even, like, what you hazily remember is going to be new for a lot of the audience that don't live in your world every day, right? so 10 years ago, you did a talk at Google Slush, talking about the democratization of AI. and, now here you are, like, open sourcing an incredible new model that we're gonna talk about. But I guess, like, what got you into democratization of AI? Like, it's not obvious from your LinkedIn or something.From Karpathy's RNN Post to SourcedEiso Kant [00:00:57]: No, it's not at all. I don't think it's obvious how I got in this space. I owe getting into this space to Andrej Karpathy.Eiso Kant [00:01:05]: In 2015, he wrote an article called “The Unreasonable Effectiveness of Recurrent Neural Nets.”Swyx [00:01:10]: Neural Nets, yep.Eiso Kant [00:01:11]: And that article, I read it, and I pivoted my startup at the time overnight to working on RNNs, and later LSTMs and Transformer models to be able to write code. If you go to this article and you scroll down, you can start seeing, like, this was the precursor to what ended up becoming language models. So, at least when he was character-level language models that were starting to predict letters, he has an example out here. There's a little Paul Graham generator, and you can read it, and the text makes sense, but it doesn't. and there's a little-- There's an example of code a little bit further down. Yeah, so Shakespeare.Swyx [00:01:47]: Shakespeare.Swyx [00:01:49]: CoolEiso Kant [00:01:49]: And for some reason, I read this, and I went down the rabbit hole of learning everything I could about RNNs and LSTMs, right? This is Transformer paper. And I had built a completely unreasonable belief, that neural nets should be able to generalize to anything and everything, and that language should be able to generalize, to a lot of things that are intelligent and the ability to write code. And so I started building Sourced, which was a fully open source company trying to build, what we used to call machine learning on code, language models on code. And we spent about four or five years on this, till the end of 2019. And that sounds really cool today, but back then, no one cared.Eiso Kant [00:02:29]: Right? Like, no one cared. We were in the dark. Like, we did things along the way. We tried applying convolutional neural nets to, like, the structure of code. We were. when attention came out, we were applying it to LSTMs, and then the Transformer paper came out. And it - it wasn't obvious, and what we missed throughout that entire journey, that we were on the right track, but we should have just kept scaling up. And today, to all of us, the scaling laws and scaling up seems like the most obvious thing. But having spent four or five years of my life on working on language models on code, it wasn't obvious. So I have a lot of respect to folks at Google and OpenAI and others who took that confidence and kept going. we failed ultimately at the time, and it was, like, biggest failure of my career, right? You blew $12 million of investors' money, which was a lot back then.Swyx [00:03:18]: Yep.Eiso Kant [00:03:19]: You spent, still a lot, but, And you spent years with, like, a group of 40 people just obsessing over this problem. And life took a different turn, And it was, and family became a focus, and I kept my heads down and really, didn't really look at language models for the following two years. big mistake considering Following years are gonna be really interesting. And then ChatGPT came out And it was like a vindication. It's like people started texting me. I found, like, my old, work decks and these old talks. And throughout that whole journey, we,ChatGPT, Vindication, and Returning to Open SourceEiso Kant [00:03:56]: We really had a strong point of view at the time that, like, as you're building more capable intelligence, it should be open and open source.Eiso Kant [00:04:04]: When we started Poolside, that wasn't the case at all, and I wanna be very open about it. When we started Poolside, we were like, there was a premise of two things. One is this technology is not gonna stop compounding in capabilities. I think to most people obvious today, but three-plus years ago when we started, most people were still arguing if these were stochastic parrots or not.Eiso Kant [00:04:23]: And the second was that reinforcement learning was gonna be the biggest driver for LLM capabilities. Today, very obvious. Three years ago, was not an opinion held or direction held at either OpenAI or Google or Anthropic or others. And so people looked down on us a little bit. They were like, “ is this really gonna work?” And so we just started working the problem, and we never really thought about open source again. We just kept our heads down and we built our, like, knowledge, understanding from scratch, right? We didn't roll out of an existing lab. So we picked up the papers and started writing code and figuring things out.Eiso Kant [00:04:59]: And it wasn't until the beginning of this year that me and my founder, Jason, picked up the open source conversation again.Eiso Kant [00:05:07]: And if you go back to some of the early things on our website, it was very straightforward. It was we wanna get to AGI, we wanna support a world of abundance, and we wanna be the first company that gets there.Eiso Kant [00:05:20]: But we started talking at the beginning of this year because it became obvious that the world was going in a direction that was starting to like, pick at us a little bit. Like, it didn't, this didn't happen overnight. It was, like, a little bit we were seeing this and we're like, “Okay, The world's going down a path.” And Throughout this journey, there was something that I used as a, as an analogy or thing. So I said well, if I go back to back in those days, 2015 or 2016, we're working on this, and I picked up a fi book off the shelf, and I was reading the book about 2035. AGI is achieved, and the story would be over the following, decades. And it would have that first chapter where everyone's trying to figure things out. You'd get the chapter of ChatGPT coming out And then you would get to the chapter where the world was at a fork in the road, and the one that it picked was one where three or four or a handful of companies were going to create all of intelligence moving forward.Eiso Kant [00:06:21]: And when I thought about that story, it felt like a dystopian fi book, not a utopian fi book. And the reality is, I'm a utopian fi guy. Like, and so We took a step back and said, “Hey, can we play a role here?” Now it was easy for us to do so because we were not at the frontier.Eiso Kant [00:06:41]: If we were at the frontier, I don't think we could have changed our mind. and I don't mean this like it's when the moment there's too much capital involved, too much expectations, you've built up things, right? We're a small team, just improving and improving. And so we knew that we could make that decision now, but it would be a lot harder to make as we got closer and closer to the frontier and caught up to others. And did a lot of soul-searching and a lot of conversations, and said, “No, this makes sense,” Even if there's big unanswered questions, like how the hell do you build a business model with foundation models about open source? Big open-ended question that we do not fully have the answer to yet, right? At what point do you no longer wanna release open source models because misuse of models has, real potential risks associated with it? how is the government gonna respond to open source? but I think it all just came down to one thing, and I'll stop the monologue, is the fact that I rather live in a world that has 100 foundation model companies than a world that has five, even if I was one of the five. And the smallest and most meaningful contribution we can make for 100 to exist is to open up our research and open up, like, our weights right now and figure out along the way how we can, like, do more.Neo-Labs, Model Choice, and the Token EconomySwyx [00:08:01]: Yeah. I think if anything, over the past three years, that has become a bit more true. you are one of a cohort of Neo labsEiso Kant [00:08:10]: YeahSwyx [00:08:10]: That people are now calling that. And, we're, we're doing this on the day that Thinky launched their, new model and you are outperforming them on their, on some benchmarks that they released, right? Like, they just don't have it yet. so it goes to show that I think, like, this is one of those things where, like, there is room for multiple players, and you are seeing a little bit more of the future. Maybe more like 20, not 100, but, like, you are one of the 20.Eiso Kant [00:08:36]: I really hope so, right? I think we I'm, I'm excited about their release, and I'm excited about everyone releasing because, like, ultimately, like, choice competition is both gonna drive progress in the right direction. But the fact that like, we create models and while we all, drink out of the same well of data effectively, we do introduce very different behaviors and biases in our models. Some are intended biases, some are completely unintended biases.Swyx [00:09:03]: Yeah.Eiso Kant [00:09:03]: And if we shape up in an ecosystem in the world where open models are gonna be a part of the token economy, like, I don't think there's any question about it anymore Then we want to be able to live in a world where companies, countries, people can choose and say, “Hey, I am most aligned and I trust most this provider for these things.”Swyx [00:09:25]: Yeah.Vibhu [00:09:26]: I think more than just one of the 20 Neo labs, up until recently, most of open source innovation was coming from the Chinese labs, right? So there's the DeepSeek of the West. Is it today? Okay, maybe it's thinking machines reflection, but there aren't many, right? So, one of the things you guys started in France, Europe, but very much now you're taking that American standpoint and more than just that, the point is the Chinese models that we see, they're not super open research. the work you put out is, I think, some of the best. So every few months you get not only frontier models, but also here's a breakdown blog, paper, technical report of here's everything for state of the art to build, frontier intelligence and you're filling that gap too, right? So not just only open weight, not just Western, but also pretty open research.Open Weights vs. Open ResearchEiso Kant [00:10:20]: No, I appreciate it. Look, I think it's, I think it's the most meaningful contribution, right? Weights are a binary. Let's call them what they are. Yes, we can modify them, we can change them, but, like, giving someone the weights does not allow them ultimately to recreate what you're doing, right? And so now there's challenges around releasing data sets, challenges around like releasing certain things, but being able to share your research, like, right, how do we do it? What are the lessons we learned that we spent, tens of thousands of experiments of compute on? I think very much so. One correction though, Vibhu, and I say this because it's been haunting us for quite a few years. We from day zero were an American company.Swyx [00:10:55]: Yeah. They movedPoolside's Global Team and American Company StorySwyx [00:10:56]: To France.Eiso Kant [00:10:56]: So the story once and for all is very. We start as an American company. We have always been an American company, and early on we made a very conscious decision. We said, “We're not gonna hire any researchers in the Bay Area. We're gonna look for talent everywhere else in the world.” and that is everything from Middle Americas, Seattle to, Serbia, and to Taiwan and Singapore and other places. And it was because we took a view that this was gonna become a talent war for this, and I think it has over the years now. Three years ago, that wasn't fully obvious yet. I think today it very much is. And we also realized that, like, some of the world's most capable people with, like, the most interesting, innovative ideas were not just gonna be here. And so it led us to create like a fully remote company. and we ended up opening an office in Paris and London and different places and we have a lot of the team in the US and a lot of team outside. But we always took this view of like, we're an American company, but if we want the best of the best to work with us, we need to take a global view. Now we do also have people here in Silicon Valley, like the company's grown and others, but I think one of the things that, it slowed us down at the beginning, but it has sped us up now, and it's why you're seeing like the progress, I think, on our models and the cadence at which we release, is because we didn't roll out of an existing lab. Right? we didn't, we didn't have a lot of the information that's freely flowing around here at the time. We just took this point of view as like, “Okay, well, let's just work the problem. Let's just go and, like, read the few papers that are out there, and let's just figure this stuff out.” And we made some hilarious mistakes in model training because of that over the yearsEiso Kant [00:12:35]: Like especially in the first 12 months. there's a few that I think still haunt me and scare me. We can talk about them later. but it created a, like, a resiliency and persistency in the team, right? with extremely few people have left us over the years, that, like, told us, “Okay, we can do this.” When we first wrote our first training code base completely from scratch, it wasn't a fork of any open source. It was just like, “Okay, let's build it from scratch.” I remember we had this one moment where we spent three weeks working out an optimizer bug. Like, it was like training just couldn't get stable. We, like, obsessed over it, and we thought, like, maybe we were wrong. Maybe we should have just forked this repo, or we should have. But then when we solved it, I still remember at the time we were like five people in the company. when we solved it, we were like, “Oh, we can do things,” like if we're just willing to work hard. and I think that culture with a very strong engineering bias has helped us, like, get to where we were. And so there's this notion of open source and talent and these things. I think we, We just took different decisions from a different starting point. and I think we are lucky. I do want to definitely call it lucky. And there was a lot of hard work at the team that now, like, that's starting to show up in results.Swyx [00:13:52]: Just ‘cause we probably won't revisit this again, but, and this is a fun recruiting challenge if someone knows the answer. What was the bug? And then we won't tell the solution, but we'An Optimizer Bug and the Value of Building From ScratchEiso Kant [00:14:01]: So the - This - You're gonna test my memory here,Swyx [00:14:04]: Oh, okayEiso Kant [00:14:04]: So but I thinkSwyx [00:14:05]: DirectlyEiso Kant [00:14:05]: I think I can recall. So if you, so if you look at, So if you take like Adam as an optimizer, you have epsilonSwyx [00:14:12]: YeahEiso Kant [00:14:13]: Which is, right, like in the denominatorSwyx [00:14:14]: Momentum and weights. YeahEiso Kant [00:14:15]: Is exactly, in the denominator. And at the time, if I recall, you looked at like the early Llama papers and things like that. People were juicing epsilon, like, quite a bit. Like, they were, like, adding, I don't know if it was E minus four or whatever, like a high value for epsilon.Eiso Kant [00:14:31]: And if you think about this during training, it's like a bit weird and counterintuitive that we're adding noise to our optimizer by just adding effectively, like, a random number in the denominator, right? Like behind the decimal point. And I don't recall the exact bug, but it had - What I remember is once we solved it, we no longer had to juice epsilon as much as, like, was happening in the Llama paper and other places. and it was like one of those fundamental moments where we had trusted this paper that was out there, and we're like, “Oh, no, it has to be this way. It has to have this high value of epsilon.” But it made no sense to us intuitively. Like, why do you have to have this so high? Like, if you're just trying to avoid division by zero, why can't the value be extremely small? and that was like one of those moments where you realize like, okay, finding things out from scratch yourself builds a better intuition. Because the one thing you learn very quickly with model building is that your intuitions that you start with are gonna get beaten up so hard.Eiso Kant [00:15:33]: Right? Like - It's such an experimental science, that the things that seem obvious, you very quickly get to learn, like, you were wrong, and hopefully you figure out why, and sometimes you don't even.Swyx [00:15:45]: Yeah. yeah, so, one of the reasons that you, when you released your new models, Vibhu got really excited. I mean, everyone got really excited. But Vibhu led our paper club on it, and you guys sawEiso Kant [00:15:58]: YeahSwyx [00:15:58]: Obviously. maybe talk through some lessons learned in that, whatever you can disclose. we can focus on the model factory stuff, whatever you think is a good starting point.Model Building as EngineeringEiso Kant [00:16:08]: So I would say that our view from very early on in the company was that model building is ultimately 90% engineering.Eiso Kant [00:16:18]: And I think we all know it in the industry because if you look at where's every researcher spending their time, they're spending their time writing code, right? Looking at data and writing code. And so we said, okay, The state at the moment, like three years ago, was bash scripts and Slurm and spaghetti code bases for training and, like, data pipelines that were patched together. And we looked at this and said, “Well, ultimately, model building is a process.” You're going from raw data, right? Like training raw material, the web, et cetera. you're doing a whole bunch of filtering, cleaning up, transformations, analyzing. These days, that's, far more complex than it was three years ago. then you're training a model, which is effectively a large distributed systems problem, right? Across hardware that has still-- It's become a lot more reliable. It was extremely flaky back then. and now with every new generation, we get our new sets of challenges. And then you go into the next stages, right? There was no training back then, but, like, you got, your post-training and then your reinforcement learning. And so we looked at this and we said, “Well, this looks like an industrialized process. This looks like an end process, that every single part of it has its machinery,” right? If it's your big data pipelines, if it's your crawling ingestion of the web, if it's your, large-scale distributed training, and then you've got your reliability. And we said, “Well, why don't we take some of the world's smartest distributed systems engineers that we knew and make them part of the process of research from day zero?” Not retrofitting it later on, but, like, really from the beginning. And that became our model factory. And so our model factory started with a handful of components. Today, it's thousands of components, and I try to equate it to, if you think about, like, someone who was at the very early days of Foxconn, if they had been there for the following, decade, they would be able to rebuild Foxconn because they saw every decision that led to building that system and all the complexity. If you and I walk into Foxconn today, no chance.The Model Factory and Experiment VelocityEiso Kant [00:18:18]: Right? Because we don't have the lineage and history of decisions that led to that. And so we built early on from the beginning- with a team that really understood that, well, the metric that we are optimizing for is the speed of an idea from a researcher to an experimental result that we can trust to then being part of the next model training.Eiso Kant [00:18:42]: And in the. And because it's such an experimental science, ultimately, in the beginning when it wasn't that complex, you could patch your way around it, right? But now, at any foundation model company, you are running. I mean, we're a small team, right? We're less than 70 researchers, another 35 engineers. and we are running, I haven't checked the latest count, but far more than 10,000, maybe 10 to 20,000 experiments a month that we cut. And so if you look at that scale of every model run that is, like it's ultimately it's, it's you need to be able to trust it as an infra problem. And so what we have now done over the years is gotten really good at that, and just by working it and improving it and obsessing over those end decisions. So now what that means is that you looked up Laguna XS 2 that we launched. It was five weeks from the beginning of training to launch. The model that we're gonna talk about today was eight weeks from start of training, to launch. We started the next model literally yesterday because we now finished the post-training required for the model we're launching, next week or by the time this comes out today. and we move that compute to the much larger Laguna M model that we're now training. And so the model should be an artifact of someone's process. It shouldn't be really a thing in itself. Like, and we treat this like the way you would look at like a SpaceX factory where, yes, the first rocket, really hard to build, but the much harder challenge was building the factory. And now they're rolling off, and no one is really thinking about the next launch anymore. So it's just another launch, it's another launch, another rocket comes off. And that's what we're trying to do with model building.Eiso Kant [00:20:22]: And what has been, which was not planned from day zero, it was in the back of our mind like this will happen one day, is that when you build a really good end model factory with really good APIs and really good engineering systems, Well, what is it perfect for? It's perfect for agents.Agents Inside the Model FactoryEiso Kant [00:20:40]: Because agents are now starting to take over more and more work in our model factory.Vibhu [00:20:43]: Yeah.Eiso Kant [00:20:44]: So I look at the screens when I walk, like when we're, we come together, in our monthly, we do monthly onsites, and I walk behind people's screens and I stop by and I talk to our researchers. And the default is all of these different agents running on their screen that are writing the code. They're launching the jobs. They're evaluating the results that are coming back from the model runs. They are, making the changes. And we're still in the driver's seat. We're still coming up with the ideas. We're still helping with the debugging. But more and more, and this is right now very profound on the data side of our pipelines in both pre and post and the synthetic data pipelines, it's starting to become more on the architecture side as well. You're starting to see these twinklings of what RSI is gonna look like.Eiso Kant [00:21:27]: And that's. So when we talk about, like to your question about our models, every talk about the model factory, And my coolest example of these things is always that when we kick off a new run, doesn't matter if it's a training like big run or if it's now a post, like one of 10 post-training versions we do for like release or many experiments, is that at any given moment, the changes that somebody made that they had experimental results from the day before make it into that run.Eiso Kant [00:21:57]: So there's not like a cutoff 90 days before. Like no, it's like literally from that moment because we can now trust the machine enough. And then you also have to invest in the reliability. So one of my favorite metrics about like Laguna S is that there was no call events, Right? Like completely zero. And we haven't had a meaningful call event, like something to wake up for, as far as I recall this entire year. now there is one asterisk to that. In usually the first six hours of launching a new model run, something breaks because you set a config wrong, you made a small mistake, et cetera. So that's usually there's a little bit of intervention, but that's always within like call periods, right? Not on call. And I think that's starting to now compound. So the model we're releasing now, I love it. It's amazing, but we're already onto the next one. and I think that's the way it should be.Laguna, Five-Week Builds, and Zero On-Call EventsVibhu [00:22:50]: Hey, I also just wanna point out, so for context, this was like a month ago. we found it in the tech report, so we just came in with, “Okay, new model's dropped. Haven't heard about it.” We wereEiso Kant [00:23:02]: Yeah, we're very used to doing this every few months.Vibhu [00:23:03]: We're, we're very much like, “ okay, look, it's like, on par with Kimi, DeepSeek, whatnot, the small ones, Gemma level. Oh, it's a very cool paper on what goes into building.” And then we hit this page, right? Like literally page two of tech report is, “This process allowed us to build the small model from scratch to delivery within five weeks applying the lessons”. And then I'm like, oh, this paper is not about here's a tech report of benchmarks and here's how many tokens it was trained on. Like for people that wanna dive more from what we're not gonna discuss on the podcast, it's all laid out here, right? FromEiso Kant [00:23:38]: YeahVibhu [00:23:39]: Custom software that agents can use to interface with training code, training data.Eiso Kant [00:23:45]: Yeah. Well, link the paper correctly, so yeah.Vibhu [00:23:47]: Yeah. All that stuff. read the paper here, but,Technical Report Principles and Streaming Training DataEiso Kant [00:23:50]: But I would like to. I love principles, and I think that is a good starting off point for maybe telling some stories. Maybe we can go one by one past the principles. I'll just call out that Dagster just got bought by a Prefect.Vibhu [00:24:01]: Yeah.Eiso Kant [00:24:01]: Isn't it fun? But yes, I'm very familiar with Dagster. just anything where like they trigger some story.Vibhu [00:24:07]: So, well, I would say, well, experiments code's obvious, but I think one of my favorite things is, I don't know where it is in here, but early on, and I still think this is the case a lot of foundation model companies, people prepare their training data sets, they get packaged up, then they get copied over to a training cluster distributed across all of the nodes, and then training starts.Vibhu [00:24:30]: And we looked at this like three years ago and we were like That makes no senseEiso Kant [00:24:36]: You lose so much time because the moment you have to rematerialize the data set, you have to make a change, you have to fix something, et cetera, you've got all this time of like repackaging it, right? Toca- tokenizing it, repacking it, moving it over to a cluster, then distributing it across the nodes. The bigger your clusters are, you start using fancy like torrent-like algorithms to like distribute your data. So why aren't we streaming data into training? Right? Something that's very common and like just basicVibhu [00:25:00]: Like just in timeEiso Kant [00:25:01]: Just in time, like good computer science like principle. And that was one of the first things that I think unlocked - the model factory. Because the moment you start thinking about, well, a training job, it doesn't matter if it's a big hero run or a small like, post-training experiment, consumes a certain number of tokens per second, right? And it's not a lot, right? From a like a data, moving data perspective. So we said, well, we have our training cluster, and then we've got like our AWS kinda setup where we can build these amazing big data pipelines. We can set things up. We use Spark underneath the hood, like all these things.Vibhu [00:25:36]: But when you say AWS, it's not actual AWS, it's your internal AWS.Eiso Kant [00:25:39]: It's our internal-- No, it's our internal like just running like our infrastructureVibhu [00:25:42]: Site web servicesEiso Kant [00:25:43]: Exactly. Our stuff running on like an AWS account or on like any hardware, right?Vibhu [00:25:47]: Yeah.Eiso Kant [00:25:48]: And so once we made that shift into I can stream data into training, all of a sudden you realize a lot of things unlock. Because now you don't have to wait for the whole data set to materialize.Immutable Data, Experiments as Code, and Scientific RigorEiso Kant [00:26:00]: You now all of a sudden when you're running data experiments about mixing data, it's a config. Because you've got these data sources that are coming in, and you just - we have this service called Blender that's in the report, where we then say, “Okay, for this run, I want 20% of this source, 10% of this source. I want this much, so many epochs of repetition. I want this to be, shuffled in a certain way,” and your training job can start while the rest of the data is even still materializing. also what it does is because all of this underneath-- So for us, we treated the data layer underneath as like an immutable data layer, and that was really important. Like experiments as code, immutable data layer means that you can always go back and understand literally down to the single token at which cursor it went in on which version of the code.Vibhu [00:26:47]: Yeah.Eiso Kant [00:26:48]: And it took us a I have to admit, like the first year of Poolside, we understood that engineering had to get great, But we didn't understand yet, that this is ultimately in support of like a good rigorous scientific progress. We were quite a - We were a very small number of people, so a lot of it was YOLO ideas and YOLO runs.Vibhu [00:27:08]: Yeah.Eiso Kant [00:27:09]: And we built great infra for the YOLO runs. But once we realized that we treated data as immutable and code as always versioned, and you could always track and trace every experiment end to end perfectly, you could repeat everything perfectly, right? You have perfect reproducibility. I can still reproduce runs from two years ago if I wanted to, right? It enables the scientific progress, like the scientific process, and I think that took us probably about a year and a half into the company to figure out. We also had some great hires, like our head of applied research, Nikolai, who joined us from Yandex, who'd been working on language models since like the early 2020s, I think brought that into the company of like, “Hey, we wanna have even more rigor.” And then once we kinda had the combination of like increasingly more capable platform that allowed people to do more, but had this immutability, we were able to start “Okay, every experiment is truly an ablation. We truly need to understand it.” And I think we became much more scientifically rigorous in the last couple of years, and the infra underneath enabled it. and then there's just fun stuff like, andVibhu [00:28:16]: Yeah, a lot of it's fun, like even just the, one, you share all the ablations, two, picking the data sets, right? There's like a random small paragraph in here where it's just like, “Oh yeah, training data, we have some, we have an auto mixer.” it trains eight small models, scales them up, picks the training data set. We don't even need to look at it. I'm like, “Wow, a lot of engineering rigor there.” And there's just, there's just a lot in here.Publishing Research and Giving BackEiso Kant [00:28:40]: Yeah, and it'- and look, and we wanna put out more. Like we, We treat writing papers as something that we haven't earned the right for yet for a long time. So you earn the right to spend time, publishing research once you're at the frontier, because until then, you're catching up, and every minute and hour in this industry matters. Like I obsess over it, not just the wall clock time from idea to result, but just general like time every day that we, waste is one that doesn't allow us to catch up. But in this case, we said, “Okay, we're gonna give ourselves.” I think we gave the team like three or four days while still doing their work, like give everything in there. And to your point earlier, if your stuff, it's easy to like put it out. And so there's so many more things that we wanna talk about over time, and we will definitely start doing. And as we earn more of the right, but also now have like added to our mission that we want more foundation model companies to exist, you'll see us like be way more proactive, and just trying to keep dropping some of those like things that we've learned along the way that can help others like speed up.Vibhu [00:29:40]: Which is the other cool side of this, right? It's, it's not like, back to your point, it's not just here's the benchmarks of our training. If you want to replicate, here's experiments of optimizers, data sets, post-training. you lay out a lot of it here alongside here's your system for how to do it? So it's, it's really like promotingEiso Kant [00:29:59]: No, thank youVibhu [00:29:59]: Other people can do the same.Eiso Kant [00:30:00]: And by the way, I also wanna make clear, right, we have been incredible-- Like we've taken a lot of advantage of the fact of all the open research that others have published, Right? And you mentioned, the Chinese labs, and we I think it's important that there's, from every country and every culture and background, including like Western companies like us, there's different models that come out that people can choose to trust. But I think we do have to give credit where credit's due, right? The incredible Chinese lab have done an amazing job at sharing their research, and we have definitely like been on the receiving end of taking advantage of that. So when you're on the receiving end of something coming to you, I think it's, you also have an obligation to give back.Swyx [00:30:39]: Do you have a favorite or underrated Chinese lab that you wanna shout out? Everyone shout outs DeepSeek.Chinese Labs, Zhipu, and PersistenceEiso Kant [00:30:44]: That's a good question.Swyx [00:30:45]: Moaan obviously for Therapsi. Yeah.Eiso Kant [00:30:48]: Yeah, look, I think, I think obviously everyone's been talking about Zhipu lately, with 5.2. I think what most people don't realize is when they started.Swyx [00:30:59]: Yeah.Eiso Kant [00:30:59]: Right? They started years before ChatGPT.Swyx [00:31:02]: They just rebranded. YeahEiso Kant [00:31:03]: And so, I've like, I remember how hard it was to work on these things Before the rest of the world got excited about it. And so I have an immense amount of respect for people, who were working on improving models when it wasn't the sexy thing to do, when believing in LLMs, was gonna get you ridiculed. I remember like back in 2016 when we were doing what we'd call, machine learning on code with some of these models. we would-- people would just laugh at us, like they'd be like, “This makes no sense. Like why are you wasting all these, like, millions of dollars on trying to figure this out?” And so I would say they're probably the one that, I think deserves a shout-out, not just because their latest model is very good, but because they fought to get here. And I think, I think every foundation model company it takes time to get here, right? It took us three years to get to the model that we're, that we're now gonna be releasing. and now the time in between the models is coming, is counted in weeks. It's no longer counted in months or years. But this stuff's hard. and if we can make it a little bit easier for the next person, like we should all do so. Because if we don't do so, we're, we've got a small window before models are really impacting recursive self-improvement to a level where catching up otherwise might become unfeasible. And we should try to, in that window, encourage as many labs or however we wanna call them, like to start. And so one of my currentEiso Kant [00:32:36]: Mission, but qualm is like I wanna encourage whoever is a researcher right now who thinks they can tackle this to go and leave and become my competitor.Eiso Kant [00:32:45]: Like start another foundation model company because I think we need it. I think otherwise we're not gonna be in the world where, I don't want to just be the fifth or the sixth company that wins. I wanna look at a world where there's lots of choice.Starting a Foundation Model CompanyVibhu [00:32:57]: What else do people not see in starting a foundation model? it's, there's a lot of compute, there's a lot of capital required, a lot of compute. You lay out model factory and how to do the training, but there's a lot there, right? That's,Eiso Kant [00:33:10]: Well, look, it's, I in turn-- this is an oversimplification, and I always asterisk it with that because it can land a little bit the wrong way in people's minds. But I think you can sum down, And I saw it, 95% of model building to just doing, you're just doing two things. You're improving data or you're improving compute efficiency. And I know that feels like an oversimplification for the incredible, like, Gifted and skilled work people do. But if you really look at it, like what are we doing? We are looking at data, we're generating new data, we're improving data. and the only way to do that is to look at the data, right? That's a big part of foundation model building. And on the other hand, we come up with these incredible breakthroughs in inference, in architecture, and new attention mechanisms. But what are they really doing? They're bringing compute efficiency. Now, we have definitely had some breakthroughs over the years that allow for more model capabilities. But at the limit, if you could train a large enough model, right, like, and you had infinite compute, we probably-- if you had infinite compute, you'd be at AGI probably already tomorrow.Eiso Kant [00:34:12]: Right? Like it's not. And so, and let me say that infinite compute with infinite ability of much faster networking because networking ends up being more of the bottleneck than compute. But, so I do think that's, those are the main things. And to just realize that this is engineering. I think it's become more obvious, but I think for quite a few years, people have held foundation model companies and researchers and others on this pedestal of like you're doing incredible magic or rocket science, or only like, Nobel laureate physicists can do this. And don't get me wrong, there are some really hard problems that need to be solved, but a lot of the work that all of us are doing on a day Is not sitting down trying to solve a math theorem. A lot of the work that we're doing is just really doing the basics right, writing good code, looking at data, improving it, running experiments, looking at plots, trying to see like, hey, trying to shape our intuitions. And a lot more people could be highly capable researchers. and I think that's, it feels far for people to do so. But I've seen in our own company, we've seen engineers become researchers because the model factory allowed them to be, have a much lower hurdle of running experiments and trying things. And one of the guys on our team who started as an engineer building our agents is a legit reinforcement learning researcher now, making real progress. and that happened in the span of like six months. that would've not been what I think most people assumed was possible, a couple of years ago.Swyx [00:35:46]: Yeah. I think one of the interesting moments is when you can self-host, like, if in a programming language, like if you can compile the language in the language, the equivalent is can you use your own tools, right? You have the pool CLI, you have your own models. presumably you're not only using your own models. There's no way. But like, what's that percentage over time?Laguna S, Persistence, and Behavioral GainsEiso Kant [00:36:10]: This is the first model that we're releasing that is starting to meaningfully contribute to our own work. It's not a it's not state-art model yet. Fable and other, they're, they're very capable models, but Laguna S Is really interesting. I'm gonna pull up the quote. Peng Ming, one of our heads of applied research, said something, last week as the model came out about 10 days ago, much better than we had hoped for or expected. And he said, I have the feeling that a lot of the gains in Laguna S come not from more intelligence, but more from different behavior, more verification, less taking things for granted, not declaring victory early, and being way more persistent. And to be honest, those are more predictive than raw intelligence for success in human also to some degree. And this was, he wrote me this on 5th of July on a Sunday, and it's been burned in my brain ever since because the Laguna S model, as you'll see it and why it does so well on benchmarks and why it does so well in using it on a day basis, is that it's just incredibly persistent. It reasons a lot. I do call that out. We have work to do on making it more efficient. We have to work to do on offering different reasoning modes. But this is the model that has been able to do things that I never thought it could do. A hundred eighteen billion 8B active model, which is not that large. It fits on a DGX Spark and still runs at, thirty, forty tokens a second on a Spark, is able to solve Erdős 397 independently. It's able to do complex programming tasks. It's able to. I asked it this morning to make me a Fi scanner without using any external libraries on my Mac, and it's, like, figuring out, like, the core WLAN API by really persistently trying to understand it without access to the internet. And more, I love vibe checking. I've probably spent eight to ten hours a day with this model for the last ten days.Eiso Kant [00:38:05]: I'm not exaggerating. I was on my eleven-hour flight yesterday. I spent ten hours reading trajectories and traces and, like, of the model.Eiso Kant [00:38:12]: And what I take away from it is exactly what Peng Ming said. We are gonna be able to squeeze so much more out of smaller models than I think we had imagined in the industry because, yes, there's intelligence and larger models are more intelligent. Like, no doubt about it. We should continue to scale up. but the behaviors of being really persistent, of being able to backtrack when you're wrong, of, like, understanding how to interact with your environment show us that we can get a lot more out of it. And this, for me, has created a bit of a Question in my mind the last couple of days. If you think about where we're using models today, right? We are using models, say, for knowledge work. Represents twenty-five percent of the global economy, twenty-five trillion dollars of work.Eiso Kant [00:39:00]: As we scale up models and they become more intelligent, we are excited about using them more and more for pushing the frontier of science.Small Models, Knowledge Work, and CommoditizationEiso Kant [00:39:08]: And if you look at the frontier of science, like true breakthroughs in science, they have been linked, they are linked to more intelligence in many places. Einstein figuring out general relativity is able to bring ideas together that other people would have not brought together. And I think one of the many dimensions of intelligence is the ability to do that, and it's something we clearly see that as models get larger and more capable, they're able to pull more ideas and threads together that a smaller model wouldn't be able to.Eiso Kant [00:39:36]: And we're starting to see examples of that in medicine and, like, in bio and other things. But if you think about the majority of knowledge work that we do, and it includes building software. I'm a software developer at heart first and foremost probably, although I probably can't say it that much anymore as I don't write production code in years, is that what makes us good is our persistence. It's our ability to encounter a problem and backtrack and say, “I need to go figure out this bug. I need to go research this. I need to go look at the documentation. I need to, like, try different, five different ways to see, like, if I can solve it.” But it is not necessarily bringing three ideas together from radically different fields. And so if we are now seeing, and I think Laguna S is an example, that we are able to make a relatively small model much more capable than I had definitely predicted or any previous, like, benchmarks had shown for any model remotely this size or even larger, At least on coding tasks, that it's because of the behaviors. And so now the question I have, and I don't have an answer, it is I know at the limit, so infinite model size, right, extremely large model, and the cost of that model is gonna be very expensive to run. We know this, right? So larger model ROI.Eiso Kant [00:40:52]: So I know that at the very limit, I'm not gonna use the world's largest model one day, quadrillion parameter, whatever crazy, like, scale we scale up, to do a basic coding task. Already today, I'm starting to size down for certain tasks.Eiso Kant [00:41:07]: So it means that there is an optimal. It means there's some curve that goes as we go up to model size for knowledge work, at some point we're at the peak, and after that, the return on investment of using a bigger model, just doesn't make sense.Eiso Kant [00:41:22]: Now, I think the question is, before I would have thought that peak was extremely very far away.Eiso Kant [00:41:30]: This model for me is the first sign that Maybe that peak is At a trillion, five trillion, ten trillion. Maybe we can just squeeze way more out of these models. I'm no longer thinking that we need two or three orders of magnitude on the largest models to be able to, solve knowledge work, the accounting, the legal, the code that we write. And so if that holds true, It is an argument for the commoditization of models. It's an argument that open source can win and, like, succeed in this world. And now it's of course a self-serving argument and it's a hopeful argument, but theoretically at the limit it works. We just have to go discover in the next couple of years of how much more we can squeeze out. Now, I do want to put a big asterisk. This does not mean I'm against scaling models. I think we ultimately only succeed if we scale our models as large as our competition. I do not like. I think we should not put our head in the sand and say we're gonna be king of open source small models. I think that's, It's a out. It's trying to be king of your own kingdom, but not realizing what the rest of the world's doing. All of us rather use a smarter, faster, more model. It's a sign of hope. And so I don't wanna overly state this is a good model. We have a long way to go to get to the state-art. But what hopefully people take away when they use this model is that the behaviors inside of it are what push it to be far more capable, less than necessarily the number of parameters.Pre-Training, Mid-Training, and RL Moving EarlierVibhu [00:43:03]: Is that mostly post-training? LikeEiso Kant [00:43:05]: YesVibhu [00:43:05]: Right.Eiso Kant [00:43:06]: It's entirely post-training.Vibhu [00:43:08]: Are we done improving anything on training? Is, like, training done?Eiso Kant [00:43:12]: No.Vibhu [00:43:12]: Okay.Eiso Kant [00:43:13]: SoVibhu [00:43:13]: I just wanted to cover training, and then we go post-trainingEiso Kant [00:43:15]: Training is not done. I mean, look, there's a part of training of just dealing with skill, right? Every new order of magnitude of model skill, you are going to get new things you gotta solve for. That'- but those are ultimately, engineering challenges.Eiso Kant [00:43:31]: I have a, I would say, a not commonly held opinion that reinforcement learning Will move earlier and earlier into training.Vibhu [00:43:42]: Yeah, training.Eiso Kant [00:43:44]: Not even training. Like training today, right, is, like if you look at - So we've been working on this for years already. and I think the best-- I think the first time we saw it out in public was the DeepSeek Zero paper. this is a year and a half ago, I think, if I recall correctly. where, you can Very early on in a model as it starts capable of being able to use language, et cetera, induce reasoning. and so the question that I have is like, we have this- we have the dataset that's the web. and the web, I think we could arguably say probably has The totality of humanity's knowledge somewhere encoded in different places. It's a huge variance degree of quality, from garbage data, and like once you look at training data, you really get humbled of like what the web is, to like, the most greatest scientific papers and best blog posts and like, best transcripts and whatnot.Eiso Kant [00:44:39]: And so now What we are trying to figure out, and have been doing a lot of work on, and it's a place where maybe not as open as we're on other things, but we will become more over time. we've been spending a couple of years really doing research on how can we turn the web into not just next token prediction, but into a way to teach the model to think earlier in its training. and I think there's a huge amount of gold to be found there. I think we are right now in, we've got some drugs in the industry. One of the drugs is distillation. Another drug is, more environments. Like, and they're great, and they make us feel good, and they make the models better, and like we're all addicted to them, and we'll use them, right? in various different ways. and but ultimately, I think we are still barely squeezing out of the web what we should be getting out of the web.Eiso Kant [00:45:33]: I think just next token prediction during training is not enough.Eiso Kant [00:45:36]: AndVibhu [00:45:38]: YeahEiso Kant [00:45:38]: I think we'll see some very interesting things still happen. and that RL in post-training to induce behaviors, to improve things, like I think - the whole world knows how to do this now. I think we're, we're scaling it up. Everyone is. But I wonder if we need to go as far as we're going today with environments. I'm not sure yetVibhu [00:46:01]: You mean we're going too far?Eiso Kant [00:46:02]: I'm, I'm not sure if the path to AGI is justVibhu [00:46:06]: Is more environmentEiso Kant [00:46:07]: More environments.Vibhu [00:46:08]: It seems like a never-ending, “Okay, I want instruction manual for this table, right? Am I gonna environment out building furniture? Or are we just gonna tail end like we need some general solution?”Eiso Kant [00:46:19]: I think there is, I think there's an ability to generalize more from the web. but I also am very encouraged, like when I look at Laguna S and, which is post-training is, well, is the big impact there. and I see like, oh, wait a second, just by making some of these behaviors much better, we're able to get so much more out of it. It just changes a little bit the way you think about intelligence.Vibhu [00:46:40]: Yeah. The analogy people draw often is the RL phase is where you don't learn as much new knowledge. You shiftEiso Kant [00:46:46]: Yeah.Vibhu [00:46:46]: Yeah. So, you shift distribution, and you can have it reason towards what you want. on your point about training, a lot of training is still just continue training in a domain, say medicine, then you do RL. So still justEiso Kant [00:47:00]: It's just better data, right? Like, I mean, training, ooh, I like how we invented this word. Like it's effectively just like,Vibhu [00:47:06]: Second phaseEiso Kant [00:47:07]: It's the second phase of training With like a really dumb way to do a curriculum. But like ultimately, what you'd want is a curriculum from token zero to token 30 whatever or 40 trillion tokens that really truly is the optimal curriculum for the model to learn. But training is essentially a stage curriculum on the web because we do not have to compute, And, effectively to try to ablate the perfect curriculum, right? And so I'm pretty sure that you'll start to see people talking soon about some other term, and there's two or - ‘cause now we do this, right? We talk stage two and stage three and stage four training and like. But ultimately, all we're doing is we're trying to assign a curriculum to the web data that we have to allow the model to learn better. I think at some point, as things get compute, as models get cheaper to run, as the next generations of compute, this will become more of a continuous spectrum. I also think the reason, by the way, you have training and like stage two and stage three is organizational, Right? It'- this is, I think, a thing where-- that we really try to avoid with the model factory is like Training exists because there's a training team now, right? There's people, or like people in training decide to focus on like a training effort. but what you really want is engineering and scale of experiments that allows for a much more continuous spectrum that you don't, you have infinite stages. Now, we're not there. Compute's not there. Organization design is not there for it yet. but I think we'll get there. we'll look back on a couple of years and be like, “Oh my God, it was so cute that we did our training data like this in such a like naïve way. Like we barely ordered it. We didn't really do a good job at likeCurriculum, Auto Research, and New ObjectivesVibhu [00:48:48]: The building that curriculum will get you that in the industry.Eiso Kant [00:48:51]: And I'll confirm that, when I talk to some researchers that this is a lot of the focus now is like how does training change and what is the next objective other than, next token prediction. I assume you don't have the answers, but you have some ideas.Vibhu [00:49:02]: We have some ideas. We're not ready to talk about it yet.Eiso Kant [00:49:05]: Yeah.Vibhu [00:49:05]: We've been working on them for years, and I think that's the one thing that's also like you asked earlier about, like what's not obvious about building a foundation model company is that you are constantly balancing the table stakes work, the recipe worksEiso Kant [00:49:19]: Yeah.Vibhu [00:49:19]: Versus like your, my crazyEiso Kant [00:49:22]: Pure researchVibhu [00:49:22]: Breakthrough.Eiso Kant [00:49:22]: Yeah.Vibhu [00:49:22]: Pure research and finding that balance and adjusting the percentage to it based on where you are in the race is really important.Eiso Kant [00:49:31]: I mean, so like, this is a nice way. I was gonna bring up auto research at some pointVibhu [00:49:35]: YesEiso Kant [00:49:35]: As another Andrej invention, or coinage, which is like, I honestly, like how many objective functions can there be, right? Like just try 1,000 of them, set it running, whatever.Vibhu [00:49:47]: Man, it's alsoEiso Kant [00:49:48]: Like what you're looking for. You're looking for loss curves like that, likeVibhu [00:49:51]: It's also a thing people take bets on, right? When you say more Neo labs, you're doing a version of we'll do foundation models, scale them up, next token predictors. A lot of other Neo labs that we see want to take a completely different approach, right? At some level, you're right. It's all, compute efficiency, and that's the net objective. But some are okay, different architecture, like vastly different amounts of compute spend. So some are different. They're not justEiso Kant [00:50:19]: YeahVibhu [00:50:19]: They're like, 99% not balancing, here's the vanilla and scale up. They're 99% on, here's novel research that'll change everything.Eiso Kant [00:50:27]: And I think, Luke, I think you. It depends when you started as well, right?Pure Research vs. Table StakesVibhu [00:50:30]: Yeah.Eiso Kant [00:50:30]: When we started, like the novel thing we did was reinforcement learning on code. No long- that's no longer novel by far, but we were like, - that's where we obsessed over when no one believed in RL. So you have to when you start the company, you have to have your own idea. You have to have something that's different that allows you to speed up, right? For us, it was RL to LLMs that later became common, like, Knowledge. But in the beginning, it wasn'tVibhu [00:50:53]: It's cool. this was like your original 2023 blogEiso Kant [00:50:57]: YeahVibhu [00:50:57]: Of purpose.Eiso Kant [00:50:58]: Yeah.Vibhu [00:50:59]: And like you do lay it all out here.Eiso Kant [00:51:01]: We laidVibhu [00:51:01]: The blog is pretty underrated, right? The whole RL on code was very early on.Eiso Kant [00:51:06]: Very early. And even we had to argue with people, like we say here things like to push beyond current capability, to train your own foundation model. We had to argue with people that it mattered that you had your own like, base model. you can fine-tune your way to success, right? major capabilities emerge from training a base model made accurate and useful during fine-tuning.Vibhu [00:51:23]: Which like, for perspective at the time, we knew closed models, OpenAI, Anthropic were huge. The open models we had were like Mistral 7B, a 30B, a 70B.Eiso Kant [00:51:35]: When weVibhu [00:51:35]: YeahEiso Kant [00:51:36]: The date on this thing is wrong. When we published this, it was April 2023. I think this was justVibhu [00:51:42]: YeahEiso Kant [00:51:42]: Happened on a migration, probably found it on archive.org.Vibhu [00:51:45]: Mistral.Eiso Kant [00:51:46]: Mistral had started, we started on the same month, right?Vibhu [00:51:49]: Yeah.Eiso Kant [00:51:49]: So this wasn't even, there was only, I think, Llama out at the timeVibhu [00:51:52]: SnellEiso Kant [00:51:52]: And that's it, right? And so, but I agree. I think we wan

Infrastructure readiness has become the real bottleneck for agentic AI in healthcare, as enterprises confront the shift from systems that generate content to systems that execute tasks across complex, regulated workflows.   In this episode, Alex Tyrrell, SVP and CTO of Health at Wolters Kluwer, examines how agentic AI changes operational demands for healthcare organizations in conversation with host Matthew DeMello, highlighting the need for domain‑adapted reasoning, granular APIs, and stronger observability as agents drive higher‑volume system interaction.   He underscores the practical implications for leaders: preparing backend systems for agent‑driven load, adapting models to real‑world workflows, and avoiding monolithic architectures that limit safe, scalable deployment.   Learn how to evaluate AI vendors by assessing leadership expertise, and why funding benchmarks can signal product maturity and stability, download our free PDF report, "5 Ways to Select the Right AI Vendor," at emerj.com/aiv1  

Spotify is shifting from a traditional audio platform into a powerful multi-format ad engine driven by advanced data targeting and transparent, value-exchange sponsorship models. Through innovative automated tools and natural language API plugins, the streaming giant is eliminating traditional production barriers so brands of all sizes can easily deploy high-ROAS campaign creative. Key Highlights

Send a text to Melissa and she'll answer it on the next episode. A pink, gritty, light-sensitive eye can derail a whole day—especially when it races through a classroom or leaps from one sibling to the next. We take the panic out of conjunctivitis by mapping the most common patterns you'll actually see at home and pairing them with precise homeopathic remedies that calm burning, reduce discharge, and help the eyes recover smoothly without harsh drops.We start by demystifying what conjunctivitis is and how it spreads, then sort the big three: viral pink eye with profuse tearing, bacterial cases with thick yellow or green discharge, and allergic flare-ups marked by relentless itch and seasonal triggers. From there, we share a clean, practical remedy guide: Euphrasia for watery, burning, light-sensitive eyes; Pulsatilla for crusted lashes and thick discharge, especially in children who crave fresh air; Apis for puffiness and stinging that loves cold compresses; Hepar sulf for painful, touch-sensitive eyes that want warmth; Argentum nitricum for stringy, rope-like mucus; Sulfur for recurrent, night-worse irritation; and Belladonna for sudden, bright-red intensity. You'll hear our simple dosing philosophy—choose one well-matched remedy and give it time—plus when to reassess and re-match as the picture shifts.We also dig into supportive care that actually helps: hygiene habits to slow the spread, warm or cool compresses for quick comfort, when to ditch contacts for glasses, and why skipping eye makeup can speed relief. We discuss susceptibility and recurrence, explaining how patterns point to smarter remedy choices and gentle prevention. By the end, you'll have a straightforward roadmap to navigate pink eye with confidence, reduce household transmission, and feel equipped to act early the next time eyes turn red and irritated.If this guide helps, subscribe, share with a friend who needs it, and leave a quick review to help others find the show. Want more hands-on learning or acute support? Visit melissacringle.com for the Inner Healing Circle and our acute consultation options.You may also gain Access to my Fullscript dispensary and save 30% by going to: https://us.fullscript.com/welcome/mcrenshaw FIND ME!

Don’t Fade and Die in AI Subscribe to our Newsletter: https://theultimatepartner.com/ebook-subscribe/ Check Out UPX: https://theultimatepartner.com/experience/ Matt Yanchyshyn, VP AWS Marketplace, Rekha Thangelapalita, Elastic GSI Leaders; Allison McFadden, Accenture AWS Leader; and James Kang of Nvidia join Ultimate Partner. In this panel discussion, leaders from Elastic, Accenture, Nvidia, and AWS dissect the urgent shifts in the ecosystem, emphasizing that partners must adapt to AI and agentic co-selling or risk fading away completely. The conversation explores the necessity of deep co-engineering, the power of multi-product solutions in the AWS marketplace, and how automated agents are now replacing traditional human sales pipeline progression. By embracing data readiness and strategic collaboration, organizations can survive the “token maxing” era, effectively scale their enterprise opportunities, and align with NVIDIA’s five-layer strategy to dominate the new cloud landscape. https://youtu.be/zUkL4Wqsa68 Key Takeaways AI agents will automate the majority of AWS partner co-selling attachments and opportunity progressions this year. Partners who fail to embrace agentic workflows and automated governance face the existential risk of fading into obsolescence. Successful multi-product offerings require a “blood to all organs” approach that benefits the client, the ISV, the GSI, and the hyperscaler simultaneously. Nvidia’s “five-layer cake” model emphasizes that successful outcomes at the application layer automatically drive growth for all underlying infrastructure. The “token maxing” phenomenon is forcing enterprises to seek cost-effective, open-model alternatives to scale their generative AI securely. Integrating GSIs and ISVs on the AWS marketplace significantly increases enterprise deal sizes and long-term customer renewal rates. If you're ready to lead through change, elevate your business, and achieve extraordinary outcomes through the power of partnership—this is your community. At Ultimate Partner® we want leaders like you to join us in the Ultimate Partner Experience – where transformation begins. Key Tags strategic collaboration agreement, data readiness engine, agentic co-sell, semantic layer, token maxing, five layer cake, accelerated computing platform, open models, cloud consumption, multi-product solutions, partner central agents, propensity data, automated opportunity progression, generative AI governance Transcript Matt Y and Panel Audio Podcast [00:00:00] Vince Menzione: You have a choice. You can embrace them and figure it out and get governance and, and make your data available. Um, use the partner, central agent, move to Agen Co-sell, or you can fade and die. [00:00:11] Vince Menzione: You can feel it happening. The ecosystem is shifting beneath us, the way Hyperscalers are partnering, how AI is remaking the channel and what it means to win in 2026. [00:00:22] Vince Menzione: Welcome to the Ultimate Partner Podcast. I’m Vince Menzi. Own your host. And each week I sit down with leaders at the intersection of technology, partnerships and outcomes. The voices shaping how ecosystems actually work. We talk about what’s real, what’s changing, and what it takes to lead in this era where the partner channel isn’t just part of the strategy. [00:00:44] Vince Menzione: It is the strategy because [00:00:46] Vince Menzione: being in the room changes everything. Let’s start. [00:00:51] Vince Menzione: We’ve got some amazing leaders joining us. So I think probably for a little bit of context, maybe just start with Rika. You can introduce yourself, your role and, uh, what, what you’ve been doing at Elastic. Yeah. [00:01:03] Rekha Thangellapalli: Yeah, sounds great. [00:01:04] Rekha Thangellapalli: Hi everyone. I’m Reka and I lead GSI Alliances at Elastic. Um, for the past 14 years, I’ve had the pleasure of building different kinds of partner ecosystems across companies such as SAP. MuleSoft, Salesforce, Coupa, and now Elastic. Um, I wanna thank Ultimate partner and Vince for having us here today. Thank you and the panel of these incredible speakers for joining me on stage. [00:01:31] Rekha Thangellapalli: Um, very excited for the conversation today. [00:01:33] Vince Menzione: We love Elastic, and you’ve had some of your other leaders on stage at other events. As such, the quality of your leadership team is amazing. Thank you. [00:01:42] Rekha Thangellapalli: I wholeheartedly agree. [00:01:45] Allison McFadden: Excellent. Um, hello everyone. Allison McFadden. I lead our North America AWS practice at Accenture. [00:01:52] Allison McFadden: Uh, I’ve been there for five years, and truth be told, it was my first partnership role, my first formal partnership role. Uh, so I can take some tips from all of you in the room here today. Prior to that, I was 21 years with IBM, and I got into partnerships because my last role at IBM was actually trying to build. [00:02:14] Allison McFadden: Linux business on the mainframe, and I had to have partners. I had to have partners to help me with workloads to run there. So I kind of learned, uh, trial by fire. But I’m excited for the conversation today. Excited to be in this room and excited to talk about what we’re doing with, uh, elastic. Thank you. [00:02:34] James Kang: Uh, my name is James Kang. Nice to see and meet everyone here. Vince, thank you for the opportunity. Thank you [00:02:38] Vince Menzione: for being here. [00:02:39] James Kang: Um, I’m with Nvidia, so I help manage the AWS partnership at Nvidia all up. Um, I guess fun fact, I’m former AWS and so I see a lot of very familiar faces here in the front row. Uh, former colleagues and then current friends. [00:02:56] James Kang: And so, uh, looking forward to the conversation. [00:02:59] Vince Menzione: Great. Well, we’ll start with an easy tia. Matt. This is not directed to you, directed to the others. So what does a successful AWS partnership look like from your C? So we’ll start with Eureka. [00:03:09] Rekha Thangellapalli: Sure. So from an ISV perspective, I think we really are looking at three things. [00:03:15] Rekha Thangellapalli: Uh, mutual investment building together. And scaling together. So when we talk about mutual investment, elastic recently signed a five-year SCA or strategic collaboration agreement with AWS. And while that is a significant milestone in our partnership, for us, what matters more is what it represents, and that is really a long-term commitment from both companies. [00:03:39] Rekha Thangellapalli: Towards product engineering, um, and joint go to market initiatives to deliver value to customers over time. And that’s what we see is that the best partnerships really compound and they build upon each other every year. Um, they don’t necessarily kind of reset every year. Um, next we talk about building together. [00:03:59] Rekha Thangellapalli: So, um. When we talk about joint solutions, we want to deliver solutions that are better together and the customers have to see us that way. And so whether it’s search, observability, or security, we’re looking at taking to market solutions that we can’t or necessarily don’t wanna take on our own. And finally we talk about scaling together. [00:04:22] Rekha Thangellapalli: And this is where marketplace, for instance, plays a big role, um, when customers can draw down on their cloud commitments, transact online and go from, you know, pilot to enterprise scale adoption in hours, not days. Um, this is when really everyone wins. Um, and this is also where partners like Accenture play a critical role. [00:04:47] Rekha Thangellapalli: Um, you know, the incredible amount of expertise that they bring, uh, the managed services capabilities and, um, their data assets actually play a huge role in having our customers realize that value faster. And, um, like Vince mentioned, at the end of the day, best partnerships are all all about creating kind of that. [00:05:07] Rekha Thangellapalli: Self-sustaining flywheel. And so it starts with investing together, building something unique, and having the customers realize that success faster because that success is really the only thing that’s gonna keep that flywheel going for everyone involved. I [00:05:26] Vince Menzione: absolutely. [00:05:26] Allison McFadden: Okay, amazing. I’m gonna riff off a few things Ika said, but from a GSI perspective. [00:05:32] Allison McFadden: A relationship with a WSA successful relationship with AWS looks slightly different. Um, so I think the first thing that we think of in the GSI Community common thread is that the client outcome and delivering value for clients is what we, what we’re striving for. Um, and so the partnership with AWS in that case, um, um, it has to, it has to. [00:06:01] Allison McFadden: Look like one team in front of our clients. So we have to show up indistinguishable, and that’s with AWS and with an ISV partner, it has to look like one solution in front of the client, especially moments that matter. So board meetings, um, you know, the time we’re gonna sign a deal, like we have to look like one team, uh, and keep our our client outcome, um, first and foremost in mind. [00:06:24] Allison McFadden: The second thing, and this is I think where the magic of all the people in this room comes into play. We can have as many discussions at a CEO level as we want. And if our client teams on the ground are not working together, it falls apart. Falls apart directly in front of the client. Yes. And that is a really hard thing to do. [00:06:45] Allison McFadden: So I’m passionate about the alliance work because that that work is what makes it happen at the corporate level. [00:06:53] James Kang: Cool. Um. I’ll start here. So in Nvidia is a accelerated computing platform company. Um, if you asked. Anyone on the, on the street about a year ago, what is ai? A lot of times they would say AI is, is open ai, or it’s philanthropic. [00:07:12] James Kang: Um, Jensen and I’ll, I’ll reference Jensen a lot today, um, because he is our leader, um, but he also sets the strategy in the direction for Nvidia. He talks a lot about AI in the metaphor of a five layer cake. And in terms of the five layer cake, you start off with the foundational bottom layer being power and energy, which sustains. [00:07:32] James Kang: All of our data centers, you move up the stack in terms of chips. So things think of Foxconn, think of TSMC. Next you have the infrastructure layer. So obvious choice is AWS, and then you get to the models where you do have the philanthropics and the open ais. But finally in at the precipice, you have the application layer. [00:07:53] James Kang: Ultimately, the reason why I mentioned all different stacks of the layers, the five layer cake, is the fact that the application layer is the most important. And so when you think about. Partners like Elastic or ServiceNow Trend, ai, CrowdStrike. Every time you pull from the application layer and you see a success, it pulls all five different components of that layer up. [00:08:13] James Kang: And so ultimately, as I think about success, it’s it’s being able to develop these co-sell wins at the application layer and really demonstrating that through extreme co-engineering and co-design with all the different application. Infrastructure, power and energy layers in mind. Um, Jensen also likes to think of himself not only as the CEO and founder, but also as the, the chief Marketing Officer. [00:08:35] James Kang: We are a very event driven company, and so at our big events like GTC or at big industry events like CES or Computex, he likes to show up on the biggest stage, biggest stages and showcase the partnerships with not only ISVs and GSIs, but also with end customers. And so that’s what I think about when I think of SA success. [00:08:56] Vince Menzione: That’s a really good point. You talked about, Allison, you talked about having an alliance strategy, or at least you teed it up, so I thought maybe we would go there for a second. Right? Like, what does a great alliance strategy look like and why is it important to the success of the partnership? [00:09:11] Allison McFadden: Man, I, uh, I have so many opinions on this. [00:09:13] Allison McFadden: We could probably be up here all day. That’s [00:09:15] Vince Menzione: okay. [00:09:16] Allison McFadden: Um, no, I think. Uh, there, there are a couple things, and the first one that comes to mind is focus. We cannot be all things to all people. Um, so when it comes to think about some of the, the work we’re doing with Elastic, we have a very, very clear point of view on what client problem we’re solving, what clients we want to talk to. [00:09:38] Allison McFadden: It helps if, um, from an ISV perspective, if there’s a very clear fit in. The Accenture portfolio or whatever, you know, SI consulting partner. You’re working with a very clear fit in the portfolio and we know what we’re not gonna go after, what we’re not gonna spend our time on because we have, we have this tendency, there’s millions of people. [00:10:00] Allison McFadden: The ecosystem chart that, you know, Vince, you showed up there, there’s so many connections. There’s probably more connections there than there are atoms in the universe, right? So, um. Defining what we do together and what we don’t do together is the first thing that pops to my mind. [00:10:19] Vince Menzione: Reka, do you have a perspective on it since we’re gonna, we’re gonna talk next about what you’ve done together, but, and I also wanna get mass perspective as a hyperscaler partner here as well. [00:10:29] Rekha Thangellapalli: Yeah, I mean from my perspective, I, I’m gonna, you know, kinda echo what Allison said is to be just maniacally focused. Yep. Um, because, especially from my perspective, so Elastic has three different solutions, right? We’ve got search, we’ve got observability, we’ve got security that map to completely different business units within Accenture. [00:10:47] Rekha Thangellapalli: And of course Accenture does a lot of things. And so, you know, when we first came together it was like. Okay, what are we gonna focus on? What industries are we gonna go after? Which segments are we gonna go after? Which customers, you know, um, outcomes are we trying to solve? And I think that sort of maniacal focus is the number one contributing factor to, to the fact that I’m like, up here on stage today. [00:11:12] Rekha Thangellapalli: Great. [00:11:14] Vince Menzione: Matt? Perspective? [00:11:16] Matt Yanchyshyn: Yeah, I, I, I guess I was trying to. To add something, uh, additional from an AWS perspective, uh, when it comes to, you know, what does a great alliance look like? Uh, AWS is obsessed with data, you know, in data we trust. And, and so the best, um, and, and this goes sales business problem, and it’s not just the engineering teams. [00:11:34] Matt Yanchyshyn: And so, uh, you know, Accenture does a good job of this elastic, definitely. And if you can come to the table with, um, quantifiable proof of the value of customer outcomes and partnerships. Um, you’ll win all the time and it’ll be a durable relationship with AWS ’cause we really are this data obsessed company and, and even the most senior sales leaders. [00:11:54] Matt Yanchyshyn: Uh, and so what I mean by that specifically is like if you, if you can show like your a RR to land an a RR conversion ratio, like in in numerical format, it’ll light up our sales leaders and, and they’ll be all, and they will co-sell with you all day long. If you can show the, I mentioned this earlier, like the AWS service, uh, whether you’re consulting company or, um, elastic and, and how the shape of customer accounts change positively when we work together. [00:12:15] Matt Yanchyshyn: That type of sort of quantifiable data works particularly well from an alliance perspective. With AWS as a partner, we, we really are like this data in sort of results out company. Um, so I, yeah, that’s just adding to the great points that were already made. I would say specific to AWS that that’s key. [00:12:30] Matt Yanchyshyn: Yeah. And I’m gonna bring up one more thing. I want to dive in on the, the joint value proposition, but you mentioned something that made a lot of sense and resonated to me about the organizations once you get out of partner, the partner world that we all know and love. Mm-hmm. Once you get down into a field organization or account management organization. [00:12:49] Matt Yanchyshyn: Not as much understanding and really organizations do a bad job here, honestly, in terms of enabling the field organizations. Do you agree? [00:12:58] Allison McFadden: I agree because I, I agree. And, um, you know, I think that’s one of the things, and, and I, I, when I joined Accenture, what we had was a lot of wicked smart architects delivering programs to clients in the field. [00:13:15] Allison McFadden: Very smart, very deep in AWS knowledge. Um, and that was awesome for the 10 clients they were staffed on and to get that understanding of how AWS works and I dream about lar, right? Like, this is a good, you know, but that takes real effort and real work. Yeah. And it’s, it’s um, almost like being a language translator. [00:13:37] Allison McFadden: Yes. For me. Yeah. So, you know, I had to deeply learn AWS so that I could. [00:13:42] Rekha Thangellapalli: Sure. [00:13:42] Allison McFadden: Teach my account teams. My account teams are really smart. They know who they’re selling to. They know their customers. They know what their customers need. They do not know what AWS has to offer always because they’ve got 20 partners lining up to try to tell their stories. [00:13:57] Allison McFadden: Um, they don’t know how to ask of the AWS team or the elastic team or the Nvidia team. Yeah. What they need [00:14:02] Vince Menzione: this co-selling piece. Yeah. [00:14:04] Allison McFadden: And so that is where, um. We had to build that muscle even around our AWS practice, which was a huge practice at Accenture, but we didn’t necessarily surround it with that kind of enablement and um, almost deal coaching layer. [00:14:21] Vince Menzione: So Elastic and Accenture came together. I dunno which one of you wants to lead this part of the conversation, but you will, right? Yeah. So tell us about the genesis of this and why. And a lot of people dunno what Elastic does, but you do some really incredible work. Like I, somebody told me one day was like, oh, you know, Uber, like, that’s elastic, powering all that. [00:14:41] Vince Menzione: Like, we don’t think about that. That the engines that you have and the, the backend to the customers, huge customers. [00:14:48] Rekha Thangellapalli: Yeah, absolutely. Um, so when AWS launched this feature last, um, reinvent where basically it allowed, you know, channel partners such as Accenture to be able to bundle up their services, their data assets with an ISV solution and put it on marketplace, um, you know, Accenture and Elastic immediately saw an opportunity. [00:15:09] Rekha Thangellapalli: Um, at the time most customers were doing gen ai. But they were running into the same challenge, which was that their data just was not ready. And by the way, this is a problem we were solving. Outside of marketplace. I think the, the feature that you guys launched just gave us a way to package it up and to be able to create this repeatable solution, which we call data readiness engine for gen ai and put it on marketplace. [00:15:40] Rekha Thangellapalli: And, um, this to me was a success because. Each company had a clear reason to invest. Um, so for Accenture, they were able to, you know, create a very differentiated services led offering. Uh, for Elastic, we were able to expand on our AI story. And for AWS, um, you know, it drives marketplace adoption, increases cloud consumption, all of that great stuff. [00:16:07] Rekha Thangellapalli: And customers, of course get. A solution to a very real problem that, that they were having. Um, and you know, the surprising part for me going through that journey was that, um. The pitching, the idea, getting the budget, getting the executive sponsorship was actually the easy part. The hard part was getting all three companies to come together, uh, to go from idea to launch in a very ambitious timeline of six weeks. [00:16:37] Rekha Thangellapalli: Nice. And so, you know, this was very much like. Doesn’t matter your title. We’re rolling up our sleeves and we are on this outcome together. Um, and so we literally built a RACI matrix, a project plan, and you know, we had daily standup calls for six weeks where literally. At least one person from each three of these companies called in, you know, got rid of any blockers and we made sure we were on target for that timeline. [00:17:07] Rekha Thangellapalli: Um, and you know, at the end we had a successful launch. But I think my favorite part about the story is the impact that we’re having and, um. My favorite story comes from a global pharmaceutical company that, you know, had basically nine petabytes of data spread across six different continents. Wow. And by working with Accenture and Elastic, they were able to build that trusted foundation that their AI and their agents can, you know, kind of safely tap into and be accessible at scale. [00:17:41] Rekha Thangellapalli: Um, so that’s my version. Allison. [00:17:44] Allison McFadden: Yeah. Well, I don’t have a lot to add. I just, I would say this is a good example of a couple of principles, right? One is having a forcing function is never a bad idea. Sign up for a big event, sign up. I’m like, I’m here with my, you know, Nvidia guys saying, sign up for the event. [00:17:58] Allison McFadden: It’ll make you move quick, right? [00:18:00] Audience Member: Yes. [00:18:00] Allison McFadden: Um, so that is one, but two, one of my mentors once told me, when you’re designing any kind of, you know, offering go to market motion, it has to get blood to all organs. If it does not get blood to all organs, it does not go [00:18:14] Vince Menzione: nice. [00:18:14] Allison McFadden: Um, [00:18:14] Vince Menzione: I love that analogy. [00:18:15] Allison McFadden: Oh, I love it. And I can talk all day. [00:18:17] Allison McFadden: That guy was brilliant. I love him. But, um, no, and, and so Elastic did a really nice job of bringing the tech to the table. Um, our team has to trust in that technology and its ability to scale, right? Um, because at Accenture we have to be able to deploy across 700,000 consultants. Um. And yeah, so I think those are the two, two things that really worked well here is we had, uh, trust in the technology solved a customer need. [00:18:50] Allison McFadden: Um, it drives, we don’t even talk about, like, yes, it drives marketplace revenue, but it unlocks work that we do that drives even more revenue to our AWS Friends. Right. So this is a, this is a, um, product that’s getting your data ready for AG agentic. It’s a messy problem that everyone’s dealing with, and it removes blockers for clients and it unlocks more, you know, ag agentic work on top of that. [00:19:15] Allison McFadden: So, blood to all organs. [00:19:17] Vince Menzione: So, was that the proposal going forward to say we need to have, we need to have trust in the solution. We need to drive significant revenue. It needs to be something all of our, you know, seven, 700,000 people. Can be a part of and help drive? Is that how you think about? [00:19:32] Allison McFadden: Yeah, and for us right now, um, it’s an interesting time for Accenture. [00:19:36] Allison McFadden: Our clients are asking a lot of us, and what it does is it having some of these accelerators helps us deliver cheaper, better, faster to our clients, which is what they’re demanding of us right now. Um, so it’s an accelerator to client outcomes. [00:19:55] Vince Menzione: James, what is NVIDIA’s role and how do, how do you enter the equation here? [00:20:00] James Kang: Yeah, it’s, um, it’s a good question. Um, I, I would say that Nvidia is probably one of the most misunderstood organizations in the world. Um, despite the, uh, the market capitalization in the valuation of the company, we have a very tiny organization. Um, what I mean by that is, um, if you think about. [00:20:20] James Kang: Salesforces and field sales organizations. Um, we’ll take Salesforce as the account or the customer. As an example, we have one account manager at NVIDIA that no, not only covers and is responsible for the relationship with Salesforce, um, but also manages. Automation Anywhere as well as DocuSign. Whereas at AWS, in contrast, like there are full armies and teams Yeah. [00:20:45] James Kang: That are supporting the Salesforce relationship. And so as you think about partnering and working with Nvidia, the focus has to be on really. Extreme co-design, but also being very prescriptive in terms of what are the very specific customer outcomes that we are solving for. And the guidance that I would give is bring in Nvidia into that equation and that conversation as early as possible because that [00:21:10] James Kang: co-engineering and co-design needs to be part of the foundational building blocks in order for you to come out with a end solution that checks all those different requirements. [00:21:20] James Kang: And so I think. Again, like going back to Nvidia, um, we like to talk about two different types of brains. A brain one and a brain two. Uh, brain One you think about the next quarter and making sure that you’re hitting the revenue targets for the next quarter. Brain two, you think about a long-term goals and potentials looking around corners and being very strategic. [00:21:41] James Kang: The saying internally is without Brain one, there is no oxygen, but without brain two, there is no future. And everyone at NVIDIA is trained to think in that brain two mentality. [00:21:52] Vince Menzione: Wow, Matt. [00:21:54] Matt Yanchyshyn: Yeah, I, I was just thinking I love the blood doll organs. Uh, and so just on, on that note, um, and, and, you know, the multi-product solutions that, that you, you built together, uh, that is a really good example of blood do organs because like we all know, that’s how customers buy. [00:22:07] Matt Yanchyshyn: They, they buy solutions and increasingly they’re looking for combinations of ISV, sometimes multiple products from multiple ISVs with services. Uh, often they’re buying it through a resell motion. You know, and they, and, and so that from a customer perspective, they want a single place to go. And so that’s the multi-product solution. [00:22:24] Matt Yanchyshyn: They wanna find everything they need, they need Accenture, they need Elastic to solve a specific solution. And I think where that’s headed is even more specific listings, like with AI powered listing experience, like, you know, elastic Plus Accenture for, I’ll make something up like a manufacturing workload. [00:22:37] Matt Yanchyshyn: And so this solution based. Uh, sort of buying is, is very customer centric. It’s what customers want. We all know that. But that’s, that’s the customer sort of organ, I guess. Um, but then, you know, you all have SCAs and those SCAs have marketplace commits. It helps if that gets transacted through marketplace helps the AWS relationship, you know that that’s an organ. [00:22:55] Matt Yanchyshyn: It’s the relationship. It’s, it’s the commercial construct and that you have, uh, that that’s another organ. You’re marketing people. They, that’s another organ. They don’t wanna land, uh, leads on a static marketing page. They wanna land a lead on a, a storefront with a multi-product solution that can actually convert and that you can actually buy it through that. [00:23:12] Matt Yanchyshyn: So the marketing person’s happy because they, they have less churn. Uh, and then, you know, our reps are happy ’cause guess how they get paid? They retire quota when they sell Marketplace. And they, we also, Jay McMain will tell you, that’s another organ called Jay or on, on you now. Um, [00:23:27] Matt Yanchyshyn: he’ll like that. I’ll call him up and tell him that. [00:23:29] Matt Yanchyshyn: Yeah, [00:23:30] Matt Yanchyshyn: but he, he’ll tell you, you know, don’t believe me. Obviously, never believe Matt, believe, believe the, the data and, and his data shows that. Those deals will close faster and larger if you use marketplace. So that’s, that’s a lot of organs. That’s the whole body. Um, but you know, when you have your customer happy ’cause that’s how they wanna buy your field happy. [00:23:45] Matt Yanchyshyn: Um, and, you know, the relationship happy and you know, your marketing team happy. Uh, and, and Jay happy. Um, and, and you know, I think that multi-product construct and, and the way you kind of use it to model a partnership and the way buyers ultimately wanna buy is, is really powerful. And so I, I think it’s, you know, it’s really a manifestation of how. [00:24:04] Matt Yanchyshyn: We kind of intend and to go to market anyway. Uh, so I think, you know, and thanks for leading the way, by the way. You’re, you’re amongst the very first, so that’s great to see. [00:24:11] Matt Yanchyshyn: So these storefronts are really helping this drive, drive this. Well, [00:24:13] Matt Yanchyshyn: that’s the next evolution. Like we’re talking about the multiproduct solution. [00:24:16] Allison McFadden: I’m JJ Accenture storefront. [00:24:17] Vince Menzione: Yeah. Oh, there you go. I mean, j and j Accenture storefront. [00:24:20] Allison McFadden: We’re gonna talk about that. [00:24:20] Matt Yanchyshyn: Yeah. I mean, [00:24:21] Matt Yanchyshyn: Accenture also leading the way yet again with storefronts. And so I think the combination of. You know, again, I was talking a lot about conversion. Yeah. And you know, buyers know sometimes they know what they wanna buy and, but if you really wanna convert that lead, you wanna land them again, something that combines, you know, elastic Accenture’s services plus software, but in a storefront that is, you know, surrounding with just the solutions they want so they don’t need to kind of go searching. [00:24:42] Matt Yanchyshyn: So, you know, ultimately reducing that time to close, I guess, really ’cause meeting the customer where they are with what they need. [00:24:51] Matt Yanchyshyn: So we talk about co-selling a little bit. We, Jay and I talk about this all the time. We gotta keep looping Jay in here, even though he is not even in town this week, but Reko, um, what does co-sell look like inside Elastic? [00:25:02] Matt Yanchyshyn: You’ve got, we talked about an incredible leadership team. I’ve gotten meet some of your leaders. Seems like you drive, you do a good job internally driving that. Let’s talk a little bit about it. [00:25:11] Rekha Thangellapalli: Yeah, and this is something I’m, I’m personally very passionate about. Um, co-sell is. Very much a journey, not a destination. [00:25:20] Rekha Thangellapalli: And I think step one for us is recognizing the different partner types that we have. Because at Elastic we work with, you know, OEMs, MSPs, resale distributors, GSIs, um, and they all bring something very unique. To the customer lifecycle and they all contribute very differently within, you know, our own sales cycle and sales process. [00:25:45] Rekha Thangellapalli: And so, you know, figuring out what is the unique benefit they bring, how do we enable them? So training and enablement is a huge piece of it, and so is making sure we’ve got the right metrics to measure success. Um, I know a lot of companies look at partner sourced as the north star, and that’s great, right? [00:26:06] Rekha Thangellapalli: Because that is undeniable. You can say, Hey, that would not exist if it wasn’t for my partner team. Um, but we’ve also noticed that when we bring in GSIs, it actually increases renewal rates. It significantly increases. Um, a RR over time. Um, it expands deal sizes and so these are very real metrics that we can point to, um, beyond just the co-sell and the partner sourced number. [00:26:32] Rekha Thangellapalli: Um, so for us it’s looking at it from a very holistic perspective, but also catering it towards that unique partner and making sure we’re doing everything we can to set them up for success and setting up the partnership for success. [00:26:47] Vince Menzione: So clo close win ratios, deal size and renewal rates? [00:26:52] Rekha Thangellapalli: Yes. For specifically for geos size. [00:26:54] Rekha Thangellapalli: Yeah. [00:26:55] Vince Menzione: Very interesting. Allison, uh, what had to change internally to produce these co-selling? We talked a little bit about the field organization and enabling a, a group of, and, you know, account sellers that are very customer focused and enabling them on the co-sell side. What had to change internally to drive that? [00:27:13] Vince Menzione: Yeah. [00:27:14] Allison McFadden: I, I might have already alluded to this a little bit in a previous answer, but, um, creating the capacity to develop, build, and sell these solutions, um, inside of a large GSI, where billable hours is kind of the number one metric on the table. Um. Is part of the investment that we had to make within Accenture to get this done? [00:27:36] Audience Member: Yeah, [00:27:36] Allison McFadden: so expert technology time. So we have technologists that understand the elastic technology. We do similar with Nvidia, by the way, we. We released some of their time to go co-develop the solution because it has to hold technical water, right? It can’t just be a marketing pitch. It can’t just be, it has to be a real, um, what’s the there, there. [00:27:59] Allison McFadden: So in order to actually do proper co-sell, we had to release some of that time. Um, to invest in those partnerships. Um, we’ve also done similar with some industry aligned business development leaders recently, so we have freed their time up to go. Uh. Open new conversations, educate client, account teams, go to clients, have conversations. [00:28:26] Allison McFadden: Um, so that, that’s a new motion that we, uh, have just kind of recently made, um, to allow them, I love this brain one, brain two also, right? So to allow them to focus on brain two, because a lot of our time. Typically spent delivery issues, you know, getting my hours, where am I charging my time? And so just freeing up a little of that capacity to do this work, um, helps get us in this brain two mode where we’re not just living to survive. [00:28:56] Vince Menzione: I. So, Matt, you’ve removed a lot. I mean, one of the things I admire, I admire AWS for being first to market and removing the most friction in marketplace of any of the vendors. Really, truly that. You talked about some of the announcements. How does some of, how does some of this tie PC central agents propensity sales plays, MCP, how does some of this tie to how, how you’re thinking about the future? [00:29:18] Vince Menzione: And how to enable more motions like this. [00:29:20] Matt Yanchyshyn: Yeah. Well, I, I think if you know my boss, UBA Borno, uh, you’ll know that she has a maniacal focus on automation. Yeah. Um, and, uh, co-sell is increasingly automated. You know, you were asking earlier about propensity data. You can get that propensity data in addition to sales plays and, uh, opportunity scores through the partner central agents. [00:29:38] Matt Yanchyshyn: So things that used to require multiple calls to A PDM, if you’re lucky to have one. Yeah. Or a p sm. Uh, you, you can now get through, through these agents, you know, uh, tech Systems, TGS, they, they manage what, over 5,500 customer opportunities with agents that they built on top of our partner Central APIs. [00:29:55] Matt Yanchyshyn: Um, and work Span has built a whole product and business that’s right on leveraging, uh, our APIs, our capabilities to sort of tie into your CRM. So, majority of all opportunities will be progressed and managed by agents. This year at AWS, we already have a majority of all customer opportunities, all app have a partner attached and I, I took a personal goal for a majority of those partner attachments, not to happen from a human. [00:30:22] Matt Yanchyshyn: But from our solution matching engine. And how do you get recommended by that solution? Matching engine, having a healthy ACE pipeline, thanks to partner central agents and the integrations you’re doing. And in addition to being the specializations and doing things like multi-product solutions and ultimately closing opportunities, you dream of LAR and so LAR will help that. [00:30:40] Allison McFadden: It’s more like a nightmare. [00:30:41] Vince Menzione: And so, you know, [00:30:42] Allison McFadden: it’s more like a nightmare, but [00:30:44] Vince Menzione: nightmare. Well, it’s, it’s, yeah. Nightmare of Laura and, and. Nice dreams of PRM, but the, um, but that’s the loop, right? I, I think, uh, increasingly co-sell for us, and in my mind, is largely a hundred percent automated. Yeah. Except for what matters most, those most largest, most strategic, most complex deals. [00:31:01] Vince Menzione: Where our highly paid and very skilled salespeople are most effectively used. [00:31:05] Vince Menzione: Yeah. [00:31:05] Vince Menzione: You know, the days of, you know, this person with 20 years experience selling, clicking, progressing opportunities through a pipeline, uh, should be over. Uh, and, and we need those people out, out selling and, and co-selling. And so that for me. [00:31:19] Vince Menzione: Yeah. That, you know, we talk a lot about co-sell, but I, I’m obsessed with automating as much of the co-sell as possible. [00:31:24] Vince Menzione: I remember going back to the ex Excel spreadsheets and, and that, that seems to be be Viva became spreadsheet jockeys. [00:31:31] Vince Menzione: Yeah. [00:31:32] Vince Menzione: And, and they stopped selling. They forgot how to sell. [00:31:34] Vince Menzione: Yeah. And people spend all this time doing lunch and learns and things like that. [00:31:36] Vince Menzione: And then, you know. Then the salespeople rotate out after 18 months and, and it, that’s, that’s the old days. Uh, you know, the new days are, are AI powered matching algorithms, uh, ag agentic co-sell, using the partner essential agents to get your data and, and putting that data to use automatically and, and what sounded like magic. [00:31:51] Vince Menzione: 12 months ago is being done, you know, by partners at massive scale across thousands of opportunities. You can do it today. And you know, I, there’s a guy named another Mike, right? Mike another Mike who they have, there’s like a guy who’s doing all this and I’m picking on Mike ’cause I, I know their system really well and I know the guy Mike grew easily built it for them. [00:32:08] Vince Menzione: Um, but, you know, I think, yeah, again, in the days of having 10 people sort of doing lunch and learn could be replaced by one or two people, building agents, uh, managing a massive pipeline. And, and that’s the future. [00:32:18] Vince Menzione: Exactly. James, your perspective on what breaks with co-selling? [00:32:22] James Kang: Oh, what breaks co-sell? Um, I would say. [00:32:25] James Kang: It, it starts and finishes with just misalignment and a loss of trust with the customer, especially when you have multiple partners or stakeholders involved. If you’re trying to do a three-way deal with a end customer and you’re not on the same page, you’re not gonna get to a successful outcome on, on the backend. [00:32:44] James Kang: Uh, the fix is a much more complicated story. I would say that to take a step back, um. We’ve talked about the five layer cake. We’ve talked about where NVIDIA kind of fits within the equation. We are invested in the ecosystem and so as different players and application organizations win and see these outcomes for end customers, we celebrate that success. [00:33:07] James Kang: Um, and as part of that kind of ethos of where NVIDIA fits within the ecosystem, we wanna make sure that not only. Our customers, but our partners like ISVs and GSIs are set up for success. Um, we do not as Nvidia sell hardware or GPUs directly to customers We use. Hyperscalers like AWS as kind of our force multiplier. [00:33:31] James Kang: And similarly we think of ISVs and GSIs as the force multipliers in terms of our extensions of how we, we kind of leverage the relationships and build the trust with our end customers. And so going back to kind of the question, Vince, I would say that it all comes back to trust and being able to build that mutual trust. [00:33:48] James Kang: Um, a lot of what we do when we co-sell with AWS is really on the software layer. Um, we actually have more software engineers at NVIDIA than we have hardware engineers, which is a weird thing to say, um, because everyone knows us for our GPUs. But because of that fact, we are heavily invested in Cuda and making sure that Cuda becomes the foundational layer for how not only our ISVs and GSIs, but also our end customers are building. [00:34:12] Vince Menzione: Very cool. So Reiki, you and James together on this production. Versus pilot with the Gentech ai. Tell us a little bit more about that. Where, where are you in the process? [00:34:24] Rekha Thangellapalli: Yeah. So I mean, in general, what we’re seeing out in the market in, in relation to sort of AI and, and customer’s journeys is that, um, at least from an elastic perspective, um, we’re seeing people very much in production when it comes to, you know, kind of AI assistant co-pilot use cases. [00:34:42] Rekha Thangellapalli: So, you know, things like, um, software development, customer support is a big one. Um, any sort of employee productivity use cases where there’s. Still a human in the loop somewhere. Um, and there’s a very like, clear path to value. And so we see the customers being in production excelling there. Um, no problem. [00:35:01] Rekha Thangellapalli: Where we’re seeing people still kind of in the pilot phase is those fully autonomous workflows where there is no human involved. The agent is reasoning on its own. Um, accessing multiple systems and taking an action on the user’s behalf. And what we’re seeing is that it’s not the intelligence of the agent that’s holding it back. [00:35:26] Rekha Thangellapalli: It’s more about giving the right context to the agent and having the right. Security kind of governance controls in place for the company to feel comfortable in putting these fully autonomous workflows into production. And that’s really the conversation we’re having is all right, what are the controls you need in place? [00:35:47] Rekha Thangellapalli: For you to release this to your business unit. Um, and what is the context that the agent is needed before we can comfortably let the agent make the decision on the user’s behalf? Um, James, I’d be interested to hear what you’re, what you’re seeing in the market [00:36:03] James Kang: plus one on all things context. I, I would even go so far as to say, um. [00:36:09] James Kang: H how many folks in the audience have heard of token maxing? Like this new term? [00:36:13] Rekha Thangellapalli: Yeah. Yeah. [00:36:14] James Kang: Um, I’ll, I’ll give a very specific example of, of Uber that went public. With the example of Claude, like they allowed all of their employees to use as many tokens as possible, and within the span of four months, they exhausted their full budget for the year, and so they had to pull back, and now there’s a cap on every employee. [00:36:33] James Kang: I think the number that’s circulating is $1,500 per month per employee, and so I think that is at least. In this multi-phase evolution of where we’re going to be and where we’re today, cost has become kind of the prohibitive force in terms of agentic AI at scale. Um, I think we are working on some very creative solutions in-house and Nvidia. [00:36:55] James Kang: Um. And we saw some really dynamic announcements this week when it comes to all things agent core, um, where we want to focus on very nimble ways for customers to be able to execute and go to market. And one extreme example of that is our investment within our open model strategy. So Nvidia, not only, again, providing GPUs, we actually offer our own op open models, which we call our Nitron models. [00:37:21] James Kang: And through our Nitron models, we are allowing customers to really develop and fine tune their own proprietary models in a cost effective manner. So right alongside the frontier models like OpenAI and Anthropic. It’s not a if then, it’s not an either or statement. It’s a, it’s a permutation, it’s an and So we’re giving you a cost effective alternative to not only bring your AgTech applications at scale by training on Nibo tron, which is open source, but then once you’ve kind of finished and fine tuned that specific training job to be able to. [00:37:53] James Kang: Go ahead and utilize your frontier models, whether it be OpenAI or Claude. And I know there’s other partners here that are providing those kind of different model capabilities. And so I think for us it’s, it’s a matter of choice. We know that this market is dynamic. It’s gonna be evolving over the next coming months as well as the next coming years. [00:38:10] James Kang: Uh, but we believe that we are positioned for a really unique dynamic expansion of AgTech use cases over the, at least the next three to six months. [00:38:20] Vince Menzione: Allison, for the partners in the room who are glazed over right now going, what do I, what do I do over the next 12 months? [00:38:26] Allison McFadden: Should I wake everybody up by saying, yeah, please. [00:38:27] Allison McFadden: Say go hurricanes. [00:38:28] Vince Menzione: Yes. [00:38:29] Allison McFadden: Is there anyone, anybody? Everyone’s like, boo. I get to leave the parade today to go home to parade. I live in Raleigh, so we’ve got our parade on Saturday. Nice. [00:38:39] Vince Menzione: Nice. [00:38:40] Allison McFadden: All right. Wake up. Um, all right. So for the $50 million partners in the room, um. $50 million is not small. You have something that works. [00:38:50] Allison McFadden: Right. This is great. What I would be thinking about is, you know, we’ve talked about focus before, but really doubling down on, you know, what is, what is your industry, what is your client like, ideal client that you serve. And build, um, almost that kind of community. You know, the, the clients we have move from firm to firm to firm. [00:39:17] Allison McFadden: And if you’ve done good work at one, you’re gonna follow ’em to the next. Um, so build that client demand in a specific place or specific client profile that is just like really knocking it out out of the park for you. Um. Scale with marketplace, right? So if you, I, I love some of the data that you were sharing in your talk earlier, um, because it’s like no overhead scaling mechanism. [00:39:45] Allison McFadden: I mean, it’s, it’s fantastic. Um, Accenture, other GSIs like us, we are investing in marketplace. So we’re investing in resources, um, to help us. Use marketplace more with our clients and we’re gonna capture, right, those storefronts. And if you’re present on marketplace, you’re gonna be able to catch, uh, yourself in that wheel. [00:40:09] Allison McFadden: So I think those are the, the kind of couple of things I would say is focus, focus, focus to drive that client demand and use scaling mechanisms like marketplace to really kind of, uh, accelerate. [00:40:24] Vince Menzione: Matt, anything to add there on the. [00:40:26] Vince Menzione: Well just, you know, Ja, James, you, I love the token maxing reference in Uber and it reminds me, you remember when cloud came out and everyone was like, oh, all these people are, are gonna use the cloud and costs are outta control and. [00:40:39] Vince Menzione: Um, a lot of people pulled back from the cloud and, and a lot of those companies no longer exist. And it’s similar with, with, uh, token maxing, like, oh, these agents are outta control. You have a choice. You can embrace them and figure it out and get governance and, and make your data available. Um, use the partner, central agent, move to agent to co-sell, or you can fade and die. [00:40:58] Vince Menzione: And, and that’s, that’s where we’re at. Uh, is, is the, the companies sitting here today embraced the cloud years ago and won. Uh, and and there’s a set of companies here today who are gonna embrace agents in the, for both buyers and sellers, and will win. And there are those who won’t and they won’t win. And so for me, it’s like we’re, we’re at a, we’re at a crossroads. [00:41:18] Vince Menzione: And, and if you’re gonna win, you gotta leap into that, you know? I love it. And, uh, and, and, and it’s, it means the cost of experimentation is so much lower now. Development and, and even business development or software development is, is agent enabled. And so you can take risks, you can experiment and, and you have to, it’s, it’s an existential moment. [00:41:37] Vince Menzione: Agreed. We’ve got a couple minutes left over for any questions. What do you think? Sure. Are there any here. I think there are a couple. Yeah, we’ve got, we’ve got a co-sell question I’m sure coming up here. [00:41:51] Audience Member: Um, I’m Cassandra, I’m the CEO of Partner Tap. And one of the questions I had was, I think, you know, the co-selling between the sellers is where things get. Really, really hard when you’re multi-partner. And so when I was listening, um, with, you know, the Accenture and Elastic together, you talked about how you had, you, you had to get these BD business development people. [00:42:22] Audience Member: Um, is this a new team that is over the client team? And how do these teams interact like with the elastic sellers? Are you doing a lot of coaching to the field and then with if AWS sellers are, are involved, like what is that whole picture? What does look like, [00:42:43] Allison McFadden: like [00:42:44] Audience Member: on the ground? I mean, that is the hardest part, I think, and that’s what we hear. [00:42:48] Allison McFadden: It’s so, it’s so, it’s so tough. Um, and I will, I’ll just say, so our business development leaders that we now have kind of. Expanded their capacity. They have always been, they have always been there. Um, but they have not been well resourced. They haven’t, they haven’t had very clear kind of job description. [00:43:12] Allison McFadden: I’m gonna say I, in the past they have been kind of focused on partner relationship. And so like more like an alliance manager and maybe working on some of the data. Right? So when I say I have nightmares about Lars, because we’re always trying to increase the LAR for Accenture and, and they were focused like in those detailed weeds of like trying to pass ACE and trying to call the PDM and all this stuff. [00:43:39] Allison McFadden: What we are doing is really pivoting them to be proper sales, business development focused on client outcomes and focused on. Technical skills to be able to describe what this solution is to the field. So, um, and because we need, I have many, many questions about, I gotta get agents to work with Eurogen co-sell so that that part somehow goes away. [00:44:05] Allison McFadden: So that’s a, that’s the thing we gotta solve still, but, um, so we’re pivoting them to be kind of driving. More of that co-sell enablement with the field, um, and taking that message to the field rather than being there, waiting for questions to come in from the field, waiting for like our field teams to discover, oh, I saw something that we’re doing with Elastic, like on a press release on LinkedIn. [00:44:30] Allison McFadden: Right. So we’re kind of trying to pivot them to be more proactive. [00:44:33] Vince Menzione: Very cool. [00:44:34] Rekha Thangellapalli: Yeah. And uh, Cassandra, that’s an excellent question because I think. Multi-party, you know, sort of tri-party offerings. The hardest part is operationalizing it at scale, right? Yeah. And so for this particular offering, we are basically having three routes to market. [00:44:51] Rekha Thangellapalli: So one is seeing how this offering fits into our existing elastic go to market. And so I am constantly enabling our field sellers to say, okay, within our three field sales place, here’s exactly where this fits in. Here are, you know, uh. Keywords that you hear in customer conversations where you bring up this offering and here’s a process of how it works. [00:45:14] Rekha Thangellapalli: Um, exactly At what sales stage do I bring in Accenture, how, you know, what are the roles and expectations? Right? So that’s on the elastic side. We’re doing the same thing on the Accenture side. So we’re doing a ton of training enablement and lunch and learns, and we’re also looking at how do we fit into. [00:45:31] Rekha Thangellapalli: Uh, Accenture’s AI transformation projects, we are the semantic layer, right, of their enterprise brain. And so it’s a whole different sales motion, um, and, you know, having the right assets, having the right process again to make sure that that goes smoothly. And then finally, we’re going directly to the customer. [00:45:49] Rekha Thangellapalli: So we are launching multiple external campaigns where, you know, if the customer raises their hand. We will, we will line up immediately. Right. Um, and so, [00:46:01] Allison McFadden: I mean, I can’t, I can’t, I can’t say how important that third leg of the stool is. ’cause the second part, she talked about getting into our catalog is the first thing. [00:46:09] Allison McFadden: ’cause my BU business development leaders have the catalog. Right. And that’s what they’re selling. So what Elastic has done has gotten into one of those offerings and then. If we have a customer that asks for it, that is the fastest way to alignment. That is like the number one thing that we respond to [00:46:26] Vince Menzione: customer at the center. [00:46:27] Vince Menzione: This is great. Well, I think we’re up to time. This was a great session. I want to thank you. This is what a great, what a great group. [00:46:34] Vince Menzione: Thanks for listening to the Ultimate Partner Podcast. If today’s conversation resonated, share it with a partner leader in your network. Subscribe where [00:46:43] Vince Menzione: you listen, and head over to the ultimate partner.com. [00:46:47] Vince Menzione: For show notes related content and the resources for this episode. And if you haven’t already, now’s the time to register for the Ultimate Partner Live Event in Reston, Virginia, October 26th through October 28th. Until next time, keep showing up in the rooms that matter because being in the room changes everything [00:47:09] I.

Send us Fan MailMost security teams keep buying tools and still feel behind. That's not because you picked the “wrong” scanner or missed the latest AI feature. It's because many organizations are trying to secure software like it's a one-time project instead of a repeatable manufacturing process. I'm joined by Darius Radford, founder and CEO of Knights Watch Cyber, to make that idea concrete and practical for any software-driven business. Darius breaks down why applications, APIs, CI/CD pipelines, cloud platforms, open source dependencies, and identity are the real battlefield now. He shares the core insight behind what he calls the Secure Software Factory: software isn't “written” anymore, it's manufactured. When you adopt a factory mindset, you build in quality control, governance, automation, standards, metrics, and feedback loops so secure software development becomes consistent instead of heroic. We also walk through the four capabilities he sees as non-negotiable: orchestrated delivery, developer-centric application security tooling, supply chain and artifact management governance, and unified risk correlation that turns endless security data into real context. We go straight at the AI era too. Developers and AI will build the next generation of products together, which makes AI security governance and testing even more urgent. Darius gives practical guardrails for AI-generated code, including input validation, session management, authentication and authorization checks, and permission modeling. We also talk about how a mature DevSecOps approach can help you win deals by proving security by design with repeatable metrics and alignment to common risk frameworks like OWASP. If you build software, lead engineering, or sell into security-conscious customers, this conversation is for you. Subscribe, share this with a builder on your team, and leave a review with your biggest question about secure software development, what are you struggling to make repeatable?Thanks for tuning in to this episode of Follow The Brand! We hope you enjoyed learning about the latest trends and strategies in Personal Branding, Business and Career Development, Financial Empowerment, Technology Innovation, and Executive Presence. To keep up with the latest insights and updates, visit 5starbdm.com.And don't miss Grant McGaugh's new book, First Light — a powerful guide to igniting your purpose and building a BRAVE brand that stands out in a changing world. - https://5starbdm.com/brave-masterclass/See you next time on Follow The Brand!

Today, we are dropping our final episode in our series The AI Control Loop, How enterprises govern the AI they've already deployed - sponsored by our friends at Wallarm.Wallarm is the AI Control Platform for Enterprise AI, protecting every AI workload, API, and application in production, giving CISOs the governance they need and CIOs the speed they demand. Organizations choose Wallarm for a complete inventory of APIs, AI agents, and AI apps, patented AI/ML-based threat detection and blocking that operates at production traffic speeds.In our final episode, we are joined by Shayne Higdon, Wallarm CEO, who closes the series by examining what the accountability moment demands from enterprise leaders, what a mature AI governance model needs to prove rather than promise, and what the next 12 to 24 months look like for organizations that get this right.QuestionsWhy is now the accountability moment for enterprise AI?What has changed between the early days of AI experimentation and today's enterprise AI deployments that makes accountability such a pressing issue?When we talk about AI accountability, what does that actually mean in practical terms? Are we talking about visibility, auditability, enforcement, ownership—or all of the above?As organizations race to deploy AI, how should CIOs balance the speed of transformation with the responsibility to govern it effectively?Why are traditional governance and security models struggling to keep pace with the way AI is being adopted across the enterprise?Given those challenges, how should boards and executive teams evaluate whether their organizations are truly ready to scale AI safely and responsibly?And once an organization believes it's ready, what does a mature AI governance model actually need to prove - not just promise?From an operational standpoint, how do capabilities like discovery, runtime monitoring, and enforcement come together to create a closed-loop approach to AI accountability?Stepping back and looking across this entire conversation, what's the one mindset shift every enterprise leader needs to make when it comes to AI security and accountability?And finally, as listeners think about what's ahead, what should they expect the future of AI security and accountability to look like over the next 6, 12, or even 24 months?Linkshttps://www.wallarm.com/https://www.linkedin.com/in/shaynehigdon/Full AbstractAbstract: Join Shayne Higdon, Wallarm CEO, for this episode, which closes the series by examining what the accountability moment demands from enterprise leaders, what a mature AI governance model needs to prove rather than promise, and what the next 12 to 24 months look like for organizations that get this right.AI deployment is not waiting for governance to catch up. Across most enterprises, the gap between how fast AI is being adopted and how well it is being governed is widening every quarter. CIOs and CISOs are not debating whether to govern AI. They are trying to figure out how, under real organizational pressure, with tools and frameworks that were built for a different threat model.That pressure is coming from every direction at once. Boards want AI transformation to move fast. Regulators want documented evidence that it is under control. Security teams want runtime visibility and enforcement capabilities that most of their current tools do not provide. And the AI systems themselves are not waiting: they are accessing data, calling external services, and making decisions continuously, in ways that after-the-fact governance cannot meaningfully constrain.This is the accountability moment. Not because the risk is new, but because the consequences of undermanaged AI are now concrete enough to land on a board agenda, an audit report, and a regulatory deadline at the same time. What accountability actually requires in practice is the full AI control loop: knowing what AI is running across the enterprise, seeing what it is doing at runtime, enforcing policy before damage compounds, and generating continuous evidence that the governance is real and not retroactive. Organizations that can demonstrate all four are in a fundamentally different position than those still assembling audit evidence from spreadsheets the week before a review.Our Sponsors:* Check out Cash App and use my code CASHAPP10 for a great deal: https://cash.app* Check out Plaud AI and use my code CODESTORY for a great deal: https://plaud.aiAdvertising Inquiries: https://redcircle.com/brandsPrivacy & Opt-Out: https://redcircle.com/privacy

Show DescriptionHow do websites handle advertising embeds, thinking DND thoughts about team topologies, does knowing UX help get jobs in web dev, using Elena for building progressive web components, and dealing with buttons in web components. Listen on WebsiteLinks BuySellAds Carbon Ads It needs to map back to a role – Eric Bailey Elena | Progressive Web Components Ariel Salminen David Darnes SponsorsNotionWrite custom tools for Notion Agents that generate assets, query live data, and hit any API. Listen for incoming webhooks from any app, then run workflows with Notion Agents, pages, databases, and external APIs. All of this, on a hosted runtime. Workers are isolated sandboxes managed by Notion, so the code behind your syncs, tools, and workflows runs on our infra instead of your servers.