Podcasts about Multimodality

In de podcastserie proefschriften spreekt aios interne geneeskunde dr. Tessa Steenbruggen met promovendi. In deze aflevering spreekt zij met Timo Kalisvaart over zijn proefschrift, getiteld: ‘Multimodality imaging in bone and soft tissue tumours: Deciphering Quantitative Data'. Timo vertelt onder andere over ontwikkelingen in beeldvormingstechnieken die onderzoek en uiteindelijk de behandeling van bot- en wekedelentumoren vooruit kunnen helpen. Timo gaat op 12 juni zijn proefschrift verdedigen aan de Universiteit van Leiden bij zijn promoteren prof. dr. Hans Bloem en prof. dr. Lioe-Fee de Geus-Oei en copromotor dr. Willem Grootjans.Referenties Inspiratietips: boek 1: De heelmeesters – Abraham Verghese; boek 2: Oorlogsparadijs – Nico Dros DCE-MRI in osteosarcoom DCE-MRI in osteosarcoom 2 Fantoom-model Validatie Fantoom-model Oxford

This week, we discuss Google being found to be a monopoly, OpenAI's “offer” to buy Chrome, and some hot takes on JSON. Plus, is it better to wait on hold or ask for a callback? Watch the YouTube Live Recording of Episode (https://www.youtube.com/watch?v=EhUxUPJv5g4) 516 (https://www.youtube.com/watch?v=EhUxUPJv5g4) Runner-up Titles Just Fine The SDT “Fine” Scale Callback Asynchronous Friendship I would love to get to know you better…over text Send you Jams to the dry cleaners. JSON Take it xslt-easy! Rundown OpenAI OpenAI in talks to pay about $3 billion to acquire AI coding startup Windsurf (https://www.cnbc.com/2025/04/16/openai-in-talks-to-pay-about-3-billion-to-acquire-startup-windsurf.html) The Cursor Mirage (https://artificialintelligencemadesimple.substack.com/p/the-cursor-mirage) AI is for Tinkerers (https://redmonk.com/kholterhoff/2023/06/27/ai-is-for-tinkerers/) Vibe Coding is for PMs (https://redmonk.com/rstephens/2025/04/18/vibe-coding-is-for-pms/) OpenAI releases new simulated reasoning models with full tool access (https://arstechnica.com/ai/2025/04/openai-releases-new-simulated-reasoning-models-with-full-tool-access/) Clouded Judgement 4.18.25 - The Hidden Value in the AI Application Layer (https://cloudedjudgement.substack.com/p/clouded-judgement-41825-the-hidden?utm_source=post-email-title&publication_id=56878&post_id=161562220&utm_campaign=email-post-title&isFreemail=true&r=2l9&triedRedirect=true&utm_medium=email) OpenAI tells judge it would buy Chrome from Google (https://www.theverge.com/news/653882/openai-chrome-google-us-judge) The Creators of Model Context Protocol (https://www.latent.space/p/mcp?utm_source=substack&utm_medium=email) Judge finds Google holds illegal online ad tech monopolies (https://www.cnbc.com/2025/04/17/judge-finds-google-holds-illegal-online-ad-tech-monopolies.html) Intuit, Owner of TurboTax, Wins Battle Against America's Taxpayers (https://prospect.org/power/2025-04-17-intuit-turbotax-wins-battle-against-taxpayers-irs-direct-file/) Relevant to your Interests Switch 2 Carts Still Taste Bad, Designed Purposefully To Be Spat Out (https://www.gamespot.com/articles/switch-2-carts-still-taste-bad-designed-purposefully-to-be-spat-out/1100-6530649/) CEO Andy Jassy's 2024 Letter to Shareholders (https://www.aboutamazon.com/news/company-news/amazon-ceo-andy-jassy-2024-letter-to-shareholders) Amazon CEO Andy Jassy says AI costs will come down (https://www.cnbc.com/2025/04/10/amazon-ceo-andy-jassys-2025-shareholder-letter.html) Happy 18th Birthday CUDA! (https://www.aboutamazon.com/news/company-news/amazon-ceo-andy-jassy-2024-letter-to-shareholders) Honeycomb Acquires Grit: A Strategic Investment in Pragmatic AI and Customer Value (https://www.honeycomb.io/blog/honeycomb-acquires-grit) Everything Announced at Google Cloud Next in 12 Minutes (https://www.youtube.com/watch?v=2OpHbyN4vEM) GitLab vs GitHub : Key Differences in 2025 (https://spacelift.io/blog/gitlab-vs-github) Old Fashioned Function Keys (https://economistwritingeveryday.com/2025/04/11/old-fashioned-function-keys/) Fake job seekers are flooding U.S. companies that are hiring for remote positions, (https://www.cnbc.com/2025/04/08/fake-job-seekers-use-ai-to-interview-for-remote-jobs-tech-ceos-say.html) NetRise raises $10M to expand software supply chain security platform (https://siliconangle.com/2025/04/15/netrise-raises-10-million-expand-software-supply-chain-security-platform/) Mark Zuckerberg's antitrust testimony aired his wildest ideas from Meta's history (https://www.theverge.com/policy/649520/zuckerberg-meta-ftc-antitrust-testimony-facebook-history) How Much Should I Be Spending On Observability? (https://www.honeycomb.io/blog/how-much-should-i-spend-on-observability-pt1) Did we just make platform engineering much easier by shipping a cloud IDP? (https://seroter.com/2025/04/16/did-we-just-make-platform-engineering-much-easier-by-shipping-a-cloud-idp/) Google Cloud Next 2025: Agentic AI Stack, Multimodality, And Sovereignty (https://www.forrester.com/blogs/google-next-2025-agentic-ai-stack-multimodality-and-sovereignty/) iPhone Shipments Down 9% in China's Q1 Smartphone Boom (https://www.macrumors.com/2025/04/18/iphone-shipments-down-in-china-q1/) Exclusive: Anthropic warns fully AI employees are a year away (https://www.axios.com/2025/04/22/ai-anthropic-virtual-employees-security) Synology requires self-branded drives for some consumer NAS systems, drops full functionality and support for third-party HDDs (https://www.tomshardware.com/pc-components/nas/synology-requires-self-branded-drives-for-some-consumer-nas-systems-drops-full-functionality-and-support-for-third-party-hdds) Porting Tailscale to Plan 9 (https://tailscale.com/blog/plan9-port?ck_subscriber_id=512840665&utm_source=convertkit&utm_medium=email&utm_campaign=[Last%20Week%20in%20AWS]%20Issue%20#418:%20Another%20New%20Capacity%20Dingus%20-%2017270009) CVE Foundation (https://www.thecvefoundation.org/) The Cursor Mirage (https://artificialintelligencemadesimple.substack.com/p/the-cursor-mirage) There's a Lot of Bad Telemetry Out There (https://blog.olly.garden/theres-a-lot-of-bad-telemetry-out-there) Gee Wiz (https://redmonk.com/rstephens/2025/04/04/gee-wiz/?ck_subscriber_id=512840665&utm_source=convertkit&utm_medium=email&utm_campaign=[Last%20Week%20in%20AWS]%20Issue%20#418:%20Another%20New%20Capacity%20Dingus%20-%2017270009) Nonsense Silicon Valley crosswalk buttons hacked to imitate Musk, Zuckerberg's voices (https://techcrunch.com/2025/04/14/silicon-valley-crosswalk-buttons-hacked-to-imitate-musk-zuckerberg-voices/) A Visit to Costco in France (https://davidlebovitz.substack.com/p/a-visit-to-costco-in-france) No sweat: Humanoid robots run a Chinese half-marathon (https://apnews.com/article/china-robot-half-marathon-153c6823bd628625106ed26267874d21) Metre, a consistent measurement of the world (https://mappingignorance.org/2025/04/23/150-years-ago-the-metre-convention-determined-how-we-measure-the-world/) Conferences DevOps Days Atlanta (https://devopsdays.org/events/2025-atlanta/welcome/), April 29th-30th. KCD Texas Austin 2025 (https://community.cncf.io/events/details/cncf-kcd-texas-presents-kcd-texas-austin-2025/), May 15th, Whitney Lee Speaking. Cloud Foundry Day US (https://events.linuxfoundation.org/cloud-foundry-day-north-america/), May 14th, Palo Alto, CA, Coté speaking. Fr (https://vmwarereg.fig-street.com/051325-tanzu-workshop/)ee AI workshop (https://vmwarereg.fig-street.com/051325-tanzu-workshop/), May 13th. day before C (https://events.linuxfoundation.org/cloud-foundry-day-north-america/)loud (https://events.linuxfoundation.org/cloud-foundry-day-north-america/) (https://events.linuxfoundation.org/cloud-foundry-day-north-america/)Foundry (https://events.linuxfoundation.org/cloud-foundry-day-north-america/) Day (https://events.linuxfoundation.org/cloud-foundry-day-north-america/). NDC Oslo (https://ndcoslo.com/), May 21st-23th, Coté speaking. SDT News & Community Join our Slack community (https://softwaredefinedtalk.slack.com/join/shared_invite/zt-1hn55iv5d-UTfN7mVX1D9D5ExRt3ZJYQ#/shared-invite/email) Email the show: questions@softwaredefinedtalk.com (mailto:questions@softwaredefinedtalk.com) Free stickers: Email your address to stickers@softwaredefinedtalk.com (mailto:stickers@softwaredefinedtalk.com) Follow us on social media: Twitter (https://twitter.com/softwaredeftalk), Threads (https://www.threads.net/@softwaredefinedtalk), Mastodon (https://hachyderm.io/@softwaredefinedtalk), LinkedIn (https://www.linkedin.com/company/software-defined-talk/), BlueSky (https://bsky.app/profile/softwaredefinedtalk.com) Watch us on: Twitch (https://www.twitch.tv/sdtpodcast), YouTube (https://www.youtube.com/channel/UCi3OJPV6h9tp-hbsGBLGsDQ/featured), Instagram (https://www.instagram.com/softwaredefinedtalk/), TikTok (https://www.tiktok.com/@softwaredefinedtalk) Book offer: Use code SDT for $20 off "Digital WTF" by Coté (https://leanpub.com/digitalwtf/c/sdt) Sponsor the show (https://www.softwaredefinedtalk.com/ads): ads@softwaredefinedtalk.com (mailto:ads@softwaredefinedtalk.com) Recommendations Brandon: Dope Thief (https://www.rottentomatoes.com/tv/dope_thief) on Apple TV (https://www.rottentomatoes.com/tv/dope_thief) Coté: Check out the recording of the Tanzu Annual update (https://www.youtube.com/watch?v=c1QZXzJcAfQ), all about Tanzu's private AI platform. Next, watch Coté's new MCP for D&D video (#4) figures out something cool to do with MCP Prompts (https://www.youtube.com/watch?v=xEtYBznneFg), they make sense now. And, a regret-a-mmendation: Fields Notes annual subscription (https://fieldnotesbrand.com/limited-editions). Photo Credits Header (https://unsplash.com/photos/a-telephone-sitting-on-top-of-a-wooden-shelf-2XnGRN_caHc)

How can AI help us understand and master deeply complex systems—from the game Go, which has 10 to the power 170 possible positions a player could pursue, or proteins, which, on average, can fold in 10 to the power 300 possible ways? This week, Reid and Aria are joined by Demis Hassabis. Demis is a British artificial intelligence researcher, co-founder, and CEO of the AI company, DeepMind. Under his leadership, DeepMind developed Alpha Go, the first AI to defeat a human world champion in Go and later created AlphaFold, which solved the 50-year-old protein folding problem. He's considered one of the most influential figures in AI. Demis, Reid, and Aria discuss game theory, medicine, multimodality, and the nature of innovation and creativity. For more info on the podcast and transcripts of all the episodes, visit https://www.possible.fm/podcast/  Listen to more from Possible here. Learn more about your ad choices. Visit podcastchoices.com/adchoices

How can AI help us understand and master deeply complex systems—from the game Go, which has 10 to the power 170 possible positions a player could pursue, or proteins, which, on average, can fold in 10 to the power 300 possible ways? This week, Reid and Aria are joined by Demis Hassabis. Demis is a British artificial intelligence researcher, co-founder, and CEO of the AI company, DeepMind. Under his leadership, DeepMind developed Alpha Go, the first AI to defeat a human world champion in Go and later created AlphaFold, which solved the 50-year-old protein folding problem. He's considered one of the most influential figures in AI. Demis, Reid, and Aria discuss game theory, medicine, multimodality, and the nature of innovation and creativity. For more info on the podcast and transcripts of all the episodes, visit https://www.possible.fm/podcast/  Select mentions:  Hitchhiker's Guide to the Galaxy by Douglas Adams AlphaGo documentary: https://www.youtube.com/watch?v=WXuK6gekU1Y Nash equilibrium & US mathematician John Forbes Nash Homo Ludens by Johan Huizinga Veo 2, an advanced, AI-powered video creation platform from Google DeepMind The Culture series by Iain Banks Hartmut Neven, German-American computer scientist Topics: 3:11 - Hellos and intros 5:20 - Brute force vs. self-learning systems 8:24 - How a learning approach helped develop new AI systems 11:29 - AlphaGo's Move 37 16:16 - What will the next Move 37 be? 19:42 - What makes an AI that can play the video game StarCraft impressive 22:32 - The importance of the act of play 26:24 - Data and synthetic data 28:33 - Midroll ad 28:39 - Is it important to have AI embedded in the world? 33:44 - The trade-off between thinking time and output quality 36:03 - Computer languages designed for AI 40:22 - The future of multimodality  43:27 - AI and geographic diversity  48:24 - AlphaFold and the future of medicine 51:18 - Rapid-fire Questions Possible is an award-winning podcast that sketches out the brightest version of the future—and what it will take to get there. Most of all, it asks: what if, in the future, everything breaks humanity's way? Tune in for grounded and speculative takes on how technology—and, in particular, AI—is inspiring change and transforming the future. Hosted by Reid Hoffman and Aria Finger, each episode features an interview with an ambitious builder or deep thinker on a topic, from art to geopolitics and from healthcare to education. These conversations also showcase another kind of guest: AI. Each episode seeks to enhance and advance our discussion about what humanity could possibly get right if we leverage technology—and our collective effort—effectively.

Send us a textIn this episode, Raúl Alberto Mora talks to us about education theory as a driver for innovative teaching, mentoring and supporting one another, and the journey of a career in Education. Raúl is known worldwide for his work in the areas of alternative literacy paradigms in second language education and research, the study of second language literacies in physical and virtual spaces, and the use of sociocritical frameworks in language education. In particular, he studies the applications of alternative literacy paradigms to analyze second-language literacy practices in urban and virtual spaces He works to understand the use of languages a social and semiotic resource. His work has been published in the Journal of Adolescent and Adult Literacy, The ALAN Review, Bilingualism and Bilingual Education, International Journal of Cultural Studies, Social Semiotics, Key Concepts in Intercultural Dialogue, Pedagogies: An International Journal, and other journals. He co-edited The Handbook of Critical Literacies, Translanguaging and Multimodality as Flow, Agency, and a New Sense of Advocacy in and From the Global South, and most recently, Reimagining Literacy in the Age of AI: Theory and Practice. Dr. Raúl Alberto Mora Velez is a researcher at the Educations, Languages, and Learning Environments research group and chairs the award-winning Literacies in Second Languages Project (LSLP) research lab. Raúl is a Research Professor at Universidad Pontificia Bolivariana in Colombia. For more information about our guest, stay tuned to the end of this episode.Links mentioned in this episode:Literacies in Second Languages Project Micro-PapersAmerican Educational Research AssociationLiteracy Research AssociationConnect with Classroom Caffeine at www.classroomcaffeine.com or on Instagram, Facebook, Twitter, and LinkedIn.

Today's episode is with Paul Klein, founder of Browserbase. We talked about building browser infrastructure for AI agents, the future of agent authentication, and their open source framework Stagehand.* [00:00:00] Introductions* [00:04:46] AI-specific challenges in browser infrastructure* [00:07:05] Multimodality in AI-Powered Browsing* [00:12:26] Running headless browsers at scale* [00:18:46] Geolocation when proxying* [00:21:25] CAPTCHAs and Agent Auth* [00:28:21] Building “User take over” functionality* [00:33:43] Stagehand: AI web browsing framework* [00:38:58] OpenAI's Operator and computer use agents* [00:44:44] Surprising use cases of Browserbase* [00:47:18] Future of browser automation and market competition* [00:53:11] Being a solo founderTranscriptAlessio [00:00:04]: Hey everyone, welcome to the Latent Space podcast. This is Alessio, partner and CTO at Decibel Partners, and I'm joined by my co-host Swyx, founder of Smol.ai.swyx [00:00:12]: Hey, and today we are very blessed to have our friends, Paul Klein, for the fourth, the fourth, CEO of Browserbase. Welcome.Paul [00:00:21]: Thanks guys. Yeah, I'm happy to be here. I've been lucky to know both of you for like a couple of years now, I think. So it's just like we're hanging out, you know, with three ginormous microphones in front of our face. It's totally normal hangout.swyx [00:00:34]: Yeah. We've actually mentioned you on the podcast, I think, more often than any other Solaris tenant. Just because like you're one of the, you know, best performing, I think, LLM tool companies that have started up in the last couple of years.Paul [00:00:50]: Yeah, I mean, it's been a whirlwind of a year, like Browserbase is actually pretty close to our first birthday. So we are one years old. And going from, you know, starting a company as a solo founder to... To, you know, having a team of 20 people, you know, a series A, but also being able to support hundreds of AI companies that are building AI applications that go out and automate the web. It's just been like, really cool. It's been happening a little too fast. I think like collectively as an AI industry, let's just take a week off together. I took my first vacation actually two weeks ago, and Operator came out on the first day, and then a week later, DeepSeat came out. And I'm like on vacation trying to chill. I'm like, we got to build with this stuff, right? So it's been a breakneck year. But I'm super happy to be here and like talk more about all the stuff we're seeing. And I'd love to hear kind of what you guys are excited about too, and share with it, you know?swyx [00:01:39]: Where to start? So people, you've done a bunch of podcasts. I think I strongly recommend Jack Bridger's Scaling DevTools, as well as Turner Novak's The Peel. And, you know, I'm sure there's others. So you covered your Twilio story in the past, talked about StreamClub, you got acquired to Mux, and then you left to start Browserbase. So maybe we just start with what is Browserbase? Yeah.Paul [00:02:02]: Browserbase is the web browser for your AI. We're building headless browser infrastructure, which are browsers that run in a server environment that's accessible to developers via APIs and SDKs. It's really hard to run a web browser in the cloud. You guys are probably running Chrome on your computers, and that's using a lot of resources, right? So if you want to run a web browser or thousands of web browsers, you can't just spin up a bunch of lambdas. You actually need to use a secure containerized environment. You have to scale it up and down. It's a stateful system. And that infrastructure is, like, super painful. And I know that firsthand, because at my last company, StreamClub, I was CTO, and I was building our own internal headless browser infrastructure. That's actually why we sold the company, is because Mux really wanted to buy our headless browser infrastructure that we'd built. And it's just a super hard problem. And I actually told my co-founders, I would never start another company unless it was a browser infrastructure company. And it turns out that's really necessary in the age of AI, when AI can actually go out and interact with websites, click on buttons, fill in forms. You need AI to do all of that work in an actual browser running somewhere on a server. And BrowserBase powers that.swyx [00:03:08]: While you're talking about it, it occurred to me, not that you're going to be acquired or anything, but it occurred to me that it would be really funny if you became the Nikita Beer of headless browser companies. You just have one trick, and you make browser companies that get acquired.Paul [00:03:23]: I truly do only have one trick. I'm screwed if it's not for headless browsers. I'm not a Go programmer. You know, I'm in AI grant. You know, browsers is an AI grant. But we were the only company in that AI grant batch that used zero dollars on AI spend. You know, we're purely an infrastructure company. So as much as people want to ask me about reinforcement learning, I might not be the best guy to talk about that. But if you want to ask about headless browser infrastructure at scale, I can talk your ear off. So that's really my area of expertise. And it's a pretty niche thing. Like, nobody has done what we're doing at scale before. So we're happy to be the experts.swyx [00:03:59]: You do have an AI thing, stagehand. We can talk about the sort of core of browser-based first, and then maybe stagehand. Yeah, stagehand is kind of the web browsing framework. Yeah.What is Browserbase? Headless Browser Infrastructure ExplainedAlessio [00:04:10]: Yeah. Yeah. And maybe how you got to browser-based and what problems you saw. So one of the first things I worked on as a software engineer was integration testing. Sauce Labs was kind of like the main thing at the time. And then we had Selenium, we had Playbrite, we had all these different browser things. But it's always been super hard to do. So obviously you've worked on this before. When you started browser-based, what were the challenges? What were the AI-specific challenges that you saw versus, there's kind of like all the usual running browser at scale in the cloud, which has been a problem for years. What are like the AI unique things that you saw that like traditional purchase just didn't cover? Yeah.AI-specific challenges in browser infrastructurePaul [00:04:46]: First and foremost, I think back to like the first thing I did as a developer, like as a kid when I was writing code, I wanted to write code that did stuff for me. You know, I wanted to write code to automate my life. And I do that probably by using curl or beautiful soup to fetch data from a web browser. And I think I still do that now that I'm in the cloud. And the other thing that I think is a huge challenge for me is that you can't just create a web site and parse that data. And we all know that now like, you know, taking HTML and plugging that into an LLM, you can extract insights, you can summarize. So it was very clear that now like dynamic web scraping became very possible with the rise of large language models or a lot easier. And that was like a clear reason why there's been more usage of headless browsers, which are necessary because a lot of modern websites don't expose all of their page content via a simple HTTP request. You know, they actually do require you to run this type of code for a specific time. JavaScript on the page to hydrate this. Airbnb is a great example. You go to airbnb.com. A lot of that content on the page isn't there until after they run the initial hydration. So you can't just scrape it with a curl. You need to have some JavaScript run. And a browser is that JavaScript engine that's going to actually run all those requests on the page. So web data retrieval was definitely one driver of starting BrowserBase and the rise of being able to summarize that within LLM. Also, I was familiar with if I wanted to automate a website, I could write one script and that would work for one website. It was very static and deterministic. But the web is non-deterministic. The web is always changing. And until we had LLMs, there was no way to write scripts that you could write once that would run on any website. That would change with the structure of the website. Click the login button. It could mean something different on many different websites. And LLMs allow us to generate code on the fly to actually control that. So I think that rise of writing the generic automation scripts that can work on many different websites, to me, made it clear that browsers are going to be a lot more useful because now you can automate a lot more things without writing. If you wanted to write a script to book a demo call on 100 websites, previously, you had to write 100 scripts. Now you write one script that uses LLMs to generate that script. That's why we built our web browsing framework, StageHand, which does a lot of that work for you. But those two things, web data collection and then enhanced automation of many different websites, it just felt like big drivers for more browser infrastructure that would be required to power these kinds of features.Alessio [00:07:05]: And was multimodality also a big thing?Paul [00:07:08]: Now you can use the LLMs to look, even though the text in the dome might not be as friendly. Maybe my hot take is I was always kind of like, I didn't think vision would be as big of a driver. For UI automation, I felt like, you know, HTML is structured text and large language models are good with structured text. But it's clear that these computer use models are often vision driven, and they've been really pushing things forward. So definitely being multimodal, like rendering the page is required to take a screenshot to give that to a computer use model to take actions on a website. And it's just another win for browser. But I'll be honest, that wasn't what I was thinking early on. I didn't even think that we'd get here so fast with multimodality. I think we're going to have to get back to multimodal and vision models.swyx [00:07:50]: This is one of those things where I forgot to mention in my intro that I'm an investor in Browserbase. And I remember that when you pitched to me, like a lot of the stuff that we have today, we like wasn't on the original conversation. But I did have my original thesis was something that we've talked about on the podcast before, which is take the GPT store, the custom GPT store, all the every single checkbox and plugin is effectively a startup. And this was the browser one. I think the main hesitation, I think I actually took a while to get back to you. The main hesitation was that there were others. Like you're not the first hit list browser startup. It's not even your first hit list browser startup. There's always a question of like, will you be the category winner in a place where there's a bunch of incumbents, to be honest, that are bigger than you? They're just not targeted at the AI space. They don't have the backing of Nat Friedman. And there's a bunch of like, you're here in Silicon Valley. They're not. I don't know.Paul [00:08:47]: I don't know if that's, that was it, but like, there was a, yeah, I mean, like, I think I tried all the other ones and I was like, really disappointed. Like my background is from working at great developer tools, companies, and nothing had like the Vercel like experience. Um, like our biggest competitor actually is partly owned by private equity and they just jacked up their prices quite a bit. And the dashboard hasn't changed in five years. And I actually used them at my last company and tried them and I was like, oh man, like there really just needs to be something that's like the experience of these great infrastructure companies, like Stripe, like clerk, like Vercel that I use in love, but oriented towards this kind of like more specific category, which is browser infrastructure, which is really technically complex. Like a lot of stuff can go wrong on the internet when you're running a browser. The internet is very vast. There's a lot of different configurations. Like there's still websites that only work with internet explorer out there. How do you handle that when you're running your own browser infrastructure? These are the problems that we have to think about and solve at BrowserBase. And it's, it's certainly a labor of love, but I built this for me, first and foremost, I know it's super cheesy and everyone says that for like their startups, but it really, truly was for me. If you look at like the talks I've done even before BrowserBase, and I'm just like really excited to try and build a category defining infrastructure company. And it's, it's rare to have a new category of infrastructure exists. We're here in the Chroma offices and like, you know, vector databases is a new category of infrastructure. Is it, is it, I mean, we can, we're in their office, so, you know, we can, we can debate that one later. That is one.Multimodality in AI-Powered Browsingswyx [00:10:16]: That's one of the industry debates.Paul [00:10:17]: I guess we go back to the LLMOS talk that Karpathy gave way long ago. And like the browser box was very clearly there and it seemed like the people who were building in this space also agreed that browsers are a core primitive of infrastructure for the LLMOS that's going to exist in the future. And nobody was building something there that I wanted to use. So I had to go build it myself.swyx [00:10:38]: Yeah. I mean, exactly that talk that, that honestly, that diagram, every box is a startup and there's the code box and then there's the. The browser box. I think at some point they will start clashing there. There's always the question of the, are you a point solution or are you the sort of all in one? And I think the point solutions tend to win quickly, but then the only ones have a very tight cohesive experience. Yeah. Let's talk about just the hard problems of browser base you have on your website, which is beautiful. Thank you. Was there an agency that you used for that? Yeah. Herb.paris.Paul [00:11:11]: They're amazing. Herb.paris. Yeah. It's H-E-R-V-E. I highly recommend for developers. Developer tools, founders to work with consumer agencies because they end up building beautiful things and the Parisians know how to build beautiful interfaces. So I got to give prep.swyx [00:11:24]: And chat apps, apparently are, they are very fast. Oh yeah. The Mistral chat. Yeah. Mistral. Yeah.Paul [00:11:31]: Late chat.swyx [00:11:31]: Late chat. And then your videos as well, it was professionally shot, right? The series A video. Yeah.Alessio [00:11:36]: Nico did the videos. He's amazing. Not the initial video that you shot at the new one. First one was Austin.Paul [00:11:41]: Another, another video pretty surprised. But yeah, I mean, like, I think when you think about how you talk about your company. You have to think about the way you present yourself. It's, you know, as a developer, you think you evaluate a company based on like the API reliability and the P 95, but a lot of developers say, is the website good? Is the message clear? Do I like trust this founder? I'm building my whole feature on. So I've tried to nail that as well as like the reliability of the infrastructure. You're right. It's very hard. And there's a lot of kind of foot guns that you run into when running headless browsers at scale. Right.Competing with Existing Headless Browser Solutionsswyx [00:12:10]: So let's pick one. You have eight features here. Seamless integration. Scalability. Fast or speed. Secure. Observable. Stealth. That's interesting. Extensible and developer first. What comes to your mind as like the top two, three hardest ones? Yeah.Running headless browsers at scalePaul [00:12:26]: I think just running headless browsers at scale is like the hardest one. And maybe can I nerd out for a second? Is that okay? I heard this is a technical audience, so I'll talk to the other nerds. Whoa. They were listening. Yeah. They're upset. They're ready. The AGI is angry. Okay. So. So how do you run a browser in the cloud? Let's start with that, right? So let's say you're using a popular browser automation framework like Puppeteer, Playwright, and Selenium. Maybe you've written a code, some code locally on your computer that opens up Google. It finds the search bar and then types in, you know, search for Latent Space and hits the search button. That script works great locally. You can see the little browser open up. You want to take that to production. You want to run the script in a cloud environment. So when your laptop is closed, your browser is doing something. The browser is doing something. Well, I, we use Amazon. You can see the little browser open up. You know, the first thing I'd reach for is probably like some sort of serverless infrastructure. I would probably try and deploy on a Lambda. But Chrome itself is too big to run on a Lambda. It's over 250 megabytes. So you can't easily start it on a Lambda. So you maybe have to use something like Lambda layers to squeeze it in there. Maybe use a different Chromium build that's lighter. And you get it on the Lambda. Great. It works. But it runs super slowly. It's because Lambdas are very like resource limited. They only run like with one vCPU. You can run one process at a time. Remember, Chromium is super beefy. It's barely running on my MacBook Air. I'm still downloading it from a pre-run. Yeah, from the test earlier, right? I'm joking. But it's big, you know? So like Lambda, it just won't work really well. Maybe it'll work, but you need something faster. Your users want something faster. Okay. Well, let's put it on a beefier instance. Let's get an EC2 server running. Let's throw Chromium on there. Great. Okay. I can, that works well with one user. But what if I want to run like 10 Chromium instances, one for each of my users? Okay. Well, I might need two EC2 instances. Maybe 10. All of a sudden, you have multiple EC2 instances. This sounds like a problem for Kubernetes and Docker, right? Now, all of a sudden, you're using ECS or EKS, the Kubernetes or container solutions by Amazon. You're spending up and down containers, and you're spending a whole engineer's time on kind of maintaining this stateful distributed system. Those are some of the worst systems to run because when it's a stateful distributed system, it means that you are bound by the connections to that thing. You have to keep the browser open while someone is working with it, right? That's just a painful architecture to run. And there's all this other little gotchas with Chromium, like Chromium, which is the open source version of Chrome, by the way. You have to install all these fonts. You want emojis working in your browsers because your vision model is looking for the emoji. You need to make sure you have the emoji fonts. You need to make sure you have all the right extensions configured, like, oh, do you want ad blocking? How do you configure that? How do you actually record all these browser sessions? Like it's a headless browser. You can't look at it. So you need to have some sort of observability. Maybe you're recording videos and storing those somewhere. It all kind of adds up to be this just giant monster piece of your project when all you wanted to do was run a lot of browsers in production for this little script to go to google.com and search. And when I see a complex distributed system, I see an opportunity to build a great infrastructure company. And we really abstract that away with Browserbase where our customers can use these existing frameworks, Playwright, Publisher, Selenium, or our own stagehand and connect to our browsers in a serverless-like way. And control them, and then just disconnect when they're done. And they don't have to think about the complex distributed system behind all of that. They just get a browser running anywhere, anytime. Really easy to connect to.swyx [00:15:55]: I'm sure you have questions. My standard question with anything, so essentially you're a serverless browser company, and there's been other serverless things that I'm familiar with in the past, serverless GPUs, serverless website hosting. That's where I come from with Netlify. One question is just like, you promised to spin up thousands of servers. You promised to spin up thousands of browsers in milliseconds. I feel like there's no real solution that does that yet. And I'm just kind of curious how. The only solution I know, which is to kind of keep a kind of warm pool of servers around, which is expensive, but maybe not so expensive because it's just CPUs. So I'm just like, you know. Yeah.Browsers as a Core Primitive in AI InfrastructurePaul [00:16:36]: You nailed it, right? I mean, how do you offer a serverless-like experience with something that is clearly not serverless, right? And the answer is, you need to be able to run... We run many browsers on single nodes. We use Kubernetes at browser base. So we have many pods that are being scheduled. We have to predictably schedule them up or down. Yes, thousands of browsers in milliseconds is the best case scenario. If you hit us with 10,000 requests, you may hit a slower cold start, right? So we've done a lot of work on predictive scaling and being able to kind of route stuff to different regions where we have multiple regions of browser base where we have different pools available. You can also pick the region you want to go to based on like lower latency, round trip, time latency. It's very important with these types of things. There's a lot of requests going over the wire. So for us, like having a VM like Firecracker powering everything under the hood allows us to be super nimble and spin things up or down really quickly with strong multi-tenancy. But in the end, this is like the complex infrastructural challenges that we have to kind of deal with at browser base. And we have a lot more stuff on our roadmap to allow customers to have more levers to pull to exchange, do you want really fast browser startup times or do you want really low costs? And if you're willing to be more flexible on that, we may be able to kind of like work better for your use cases.swyx [00:17:44]: Since you used Firecracker, shouldn't Fargate do that for you or did you have to go lower level than that? We had to go lower level than that.Paul [00:17:51]: I find this a lot with Fargate customers, which is alarming for Fargate. We used to be a giant Fargate customer. Actually, the first version of browser base was ECS and Fargate. And unfortunately, it's a great product. I think we were actually the largest Fargate customer in our region for a little while. No, what? Yeah, seriously. And unfortunately, it's a great product, but I think if you're an infrastructure company, you actually have to have a deeper level of control over these primitives. I think it's the same thing is true with databases. We've used other database providers and I think-swyx [00:18:21]: Yeah, serverless Postgres.Paul [00:18:23]: Shocker. When you're an infrastructure company, you're on the hook if any provider has an outage. And I can't tell my customers like, hey, we went down because so-and-so went down. That's not acceptable. So for us, we've really moved to bringing things internally. It's kind of opposite of what we preach. We tell our customers, don't build this in-house, but then we're like, we build a lot of stuff in-house. But I think it just really depends on what is in the critical path. We try and have deep ownership of that.Alessio [00:18:46]: On the distributed location side, how does that work for the web where you might get sort of different content in different locations, but the customer is expecting, you know, if you're in the US, I'm expecting the US version. But if you're spinning up my browser in France, I might get the French version. Yeah.Paul [00:19:02]: Yeah. That's a good question. Well, generally, like on the localization, there is a thing called locale in the browser. You can set like what your locale is. If you're like in the ENUS browser or not, but some things do IP, IP based routing. And in that case, you may want to have a proxy. Like let's say you're running something in the, in Europe, but you want to make sure you're showing up from the US. You may want to use one of our proxy features so you can turn on proxies to say like, make sure these connections always come from the United States, which is necessary too, because when you're browsing the web, you're coming from like a, you know, data center IP, and that can make things a lot harder to browse web. So we do have kind of like this proxy super network. Yeah. We have a proxy for you based on where you're going, so you can reliably automate the web. But if you get scheduled in Europe, that doesn't happen as much. We try and schedule you as close to, you know, your origin that you're trying to go to. But generally you have control over the regions you can put your browsers in. So you can specify West one or East one or Europe. We only have one region of Europe right now, actually. Yeah.Alessio [00:19:55]: What's harder, the browser or the proxy? I feel like to me, it feels like actually proxying reliably at scale. It's much harder than spending up browsers at scale. I'm curious. It's all hard.Paul [00:20:06]: It's layers of hard, right? Yeah. I think it's different levels of hard. I think the thing with the proxy infrastructure is that we work with many different web proxy providers and some are better than others. Some have good days, some have bad days. And our customers who've built browser infrastructure on their own, they have to go and deal with sketchy actors. Like first they figure out their own browser infrastructure and then they got to go buy a proxy. And it's like you can pay in Bitcoin and it just kind of feels a little sus, right? It's like you're buying drugs when you're trying to get a proxy online. We have like deep relationships with these counterparties. We're able to audit them and say, is this proxy being sourced ethically? Like it's not running on someone's TV somewhere. Is it free range? Yeah. Free range organic proxies, right? Right. We do a level of diligence. We're SOC 2. So we have to understand what is going on here. But then we're able to make sure that like we route around proxy providers not working. There's proxy providers who will just, the proxy will stop working all of a sudden. And then if you don't have redundant proxying on your own browsers, that's hard down for you or you may get some serious impacts there. With us, like we intelligently know, hey, this proxy is not working. Let's go to this one. And you can kind of build a network of multiple providers to really guarantee the best uptime for our customers. Yeah. So you don't own any proxies? We don't own any proxies. You're right. The team has been saying who wants to like take home a little proxy server, but not yet. We're not there yet. You know?swyx [00:21:25]: It's a very mature market. I don't think you should build that yourself. Like you should just be a super customer of them. Yeah. Scraping, I think, is the main use case for that. I guess. Well, that leads us into CAPTCHAs and also off, but let's talk about CAPTCHAs. You had a little spiel that you wanted to talk about CAPTCHA stuff.Challenges of Scaling Browser InfrastructurePaul [00:21:43]: Oh, yeah. I was just, I think a lot of people ask, if you're thinking about proxies, you're thinking about CAPTCHAs too. I think it's the same thing. You can go buy CAPTCHA solvers online, but it's the same buying experience. It's some sketchy website, you have to integrate it. It's not fun to buy these things and you can't really trust that the docs are bad. What Browserbase does is we integrate a bunch of different CAPTCHAs. We do some stuff in-house, but generally we just integrate with a bunch of known vendors and continually monitor and maintain these things and say, is this working or not? Can we route around it or not? These are CAPTCHA solvers. CAPTCHA solvers, yeah. Not CAPTCHA providers, CAPTCHA solvers. Yeah, sorry. CAPTCHA solvers. We really try and make sure all of that works for you. I think as a dev, if I'm buying infrastructure, I want it all to work all the time and it's important for us to provide that experience by making sure everything does work and monitoring it on our own. Yeah. Right now, the world of CAPTCHAs is tricky. I think AI agents in particular are very much ahead of the internet infrastructure. CAPTCHAs are designed to block all types of bots, but there are now good bots and bad bots. I think in the future, CAPTCHAs will be able to identify who a good bot is, hopefully via some sort of KYC. For us, we've been very lucky. We have very little to no known abuse of Browserbase because we really look into who we work with. And for certain types of CAPTCHA solving, we only allow them on certain types of plans because we want to make sure that we can know what people are doing, what their use cases are. And that's really allowed us to try and be an arbiter of good bots, which is our long term goal. I want to build great relationships with people like Cloudflare so we can agree, hey, here are these acceptable bots. We'll identify them for you and make sure we flag when they come to your website. This is a good bot, you know?Alessio [00:23:23]: I see. And Cloudflare said they want to do more of this. So they're going to set by default, if they think you're an AI bot, they're going to reject. I'm curious if you think this is something that is going to be at the browser level or I mean, the DNS level with Cloudflare seems more where it should belong. But I'm curious how you think about it.Paul [00:23:40]: I think the web's going to change. You know, I think that the Internet as we have it right now is going to change. And we all need to just accept that the cat is out of the bag. And instead of kind of like wishing the Internet was like it was in the 2000s, we can have free content line that wouldn't be scraped. It's just it's not going to happen. And instead, we should think about like, one, how can we change? How can we change the models of, you know, information being published online so people can adequately commercialize it? But two, how do we rebuild applications that expect that AI agents are going to log in on their behalf? Those are the things that are going to allow us to kind of like identify good and bad bots. And I think the team at Clerk has been doing a really good job with this on the authentication side. I actually think that auth is the biggest thing that will prevent agents from accessing stuff, not captchas. And I think there will be agent auth in the future. I don't know if it's going to happen from an individual company, but actually authentication providers that have a, you know, hidden login as agent feature, which will then you put in your email, you'll get a push notification, say like, hey, your browser-based agent wants to log into your Airbnb. You can approve that and then the agent can proceed. That really circumvents the need for captchas or logging in as you and sharing your password. I think agent auth is going to be one way we identify good bots going forward. And I think a lot of this captcha solving stuff is really short-term problems as the internet kind of reorients itself around how it's going to work with agents browsing the web, just like people do. Yeah.Managing Distributed Browser Locations and Proxiesswyx [00:24:59]: Stitch recently was on Hacker News for talking about agent experience, AX, which is a thing that Netlify is also trying to clone and coin and talk about. And we've talked about this on our previous episodes before in a sense that I actually think that's like maybe the only part of the tech stack that needs to be kind of reinvented for agents. Everything else can stay the same, CLIs, APIs, whatever. But auth, yeah, we need agent auth. And it's mostly like short-lived, like it should not, it should be a distinct, identity from the human, but paired. I almost think like in the same way that every social network should have your main profile and then your alt accounts or your Finsta, it's almost like, you know, every, every human token should be paired with the agent token and the agent token can go and do stuff on behalf of the human token, but not be presumed to be the human. Yeah.Paul [00:25:48]: It's like, it's, it's actually very similar to OAuth is what I'm thinking. And, you know, Thread from Stitch is an investor, Colin from Clerk, Octaventures, all investors in browser-based because like, I hope they solve this because they'll make browser-based submission more possible. So we don't have to overcome all these hurdles, but I think it will be an OAuth-like flow where an agent will ask to log in as you, you'll approve the scopes. Like it can book an apartment on Airbnb, but it can't like message anybody. And then, you know, the agent will have some sort of like role-based access control within an application. Yeah. I'm excited for that.swyx [00:26:16]: The tricky part is just, there's one, one layer of delegation here, which is like, you're authoring my user's user or something like that. I don't know if that's tricky or not. Does that make sense? Yeah.Paul [00:26:25]: You know, actually at Twilio, I worked on the login identity and access. Management teams, right? So like I built Twilio's login page.swyx [00:26:31]: You were an intern on that team and then you became the lead in two years? Yeah.Paul [00:26:34]: Yeah. I started as an intern in 2016 and then I was the tech lead of that team. How? That's not normal. I didn't have a life. He's not normal. Look at this guy. I didn't have a girlfriend. I just loved my job. I don't know. I applied to 500 internships for my first job and I got rejected from every single one of them except for Twilio and then eventually Amazon. And they took a shot on me and like, I was getting paid money to write code, which was my dream. Yeah. Yeah. I'm very lucky that like this coding thing worked out because I was going to be doing it regardless. And yeah, I was able to kind of spend a lot of time on a team that was growing at a company that was growing. So it informed a lot of this stuff here. I think these are problems that have been solved with like the SAML protocol with SSO. I think it's a really interesting stuff with like WebAuthn, like these different types of authentication, like schemes that you can use to authenticate people. The tooling is all there. It just needs to be tweaked a little bit to work for agents. And I think the fact that there are companies that are already. Providing authentication as a service really sets it up. Well, the thing that's hard is like reinventing the internet for agents. We don't want to rebuild the internet. That's an impossible task. And I think people often say like, well, we'll have this second layer of APIs built for agents. I'm like, we will for the top use cases, but instead of we can just tweak the internet as is, which is on the authentication side, I think we're going to be the dumb ones going forward. Unfortunately, I think AI is going to be able to do a lot of the tasks that we do online, which means that it will be able to go to websites, click buttons on our behalf and log in on our behalf too. So with this kind of like web agent future happening, I think with some small structural changes, like you said, it feels like it could all slot in really nicely with the existing internet.Handling CAPTCHAs and Agent Authenticationswyx [00:28:08]: There's one more thing, which is the, your live view iframe, which lets you take, take control. Yeah. Obviously very key for operator now, but like, was, is there anything interesting technically there or that the people like, well, people always want this.Paul [00:28:21]: It was really hard to build, you know, like, so, okay. Headless browsers, you don't see them, right. They're running. They're running in a cloud somewhere. You can't like look at them. And I just want to really make, it's a weird name. I wish we came up with a better name for this thing, but you can't see them. Right. But customers don't trust AI agents, right. At least the first pass. So what we do with our live view is that, you know, when you use browser base, you can actually embed a live view of the browser running in the cloud for your customer to see it working. And that's what the first reason is the build trust, like, okay, so I have this script. That's going to go automate a website. I can embed it into my web application via an iframe and my customer can watch. I think. And then we added two way communication. So now not only can you watch the browser kind of being operated by AI, if you want to pause and actually click around type within this iframe that's controlling a browser, that's also possible. And this is all thanks to some of the lower level protocol, which is called the Chrome DevTools protocol. It has a API called start screencast, and you can also send mouse clicks and button clicks to a remote browser. And this is all embeddable within iframes. You have a browser within a browser, yo. And then you simulate the screen, the click on the other side. Exactly. And this is really nice often for, like, let's say, a capture that can't be solved. You saw this with Operator, you know, Operator actually uses a different approach. They use VNC. So, you know, you're able to see, like, you're seeing the whole window here. What we're doing is something a little lower level with the Chrome DevTools protocol. It's just PNGs being streamed over the wire. But the same thing is true, right? Like, hey, I'm running a window. Pause. Can you do something in this window? Human. Okay, great. Resume. Like sometimes 2FA tokens. Like if you get that text message, you might need a person to type that in. Web agents need human-in-the-loop type workflows still. You still need a person to interact with the browser. And building a UI to proxy that is kind of hard. You may as well just show them the whole browser and say, hey, can you finish this up for me? And then let the AI proceed on afterwards. Is there a future where I stream my current desktop to browser base? I don't think so. I think we're very much cloud infrastructure. Yeah. You know, but I think a lot of the stuff we're doing, we do want to, like, build tools. Like, you know, we'll talk about the stage and, you know, web agent framework in a second. But, like, there's a case where a lot of people are going desktop first for, you know, consumer use. And I think cloud is doing a lot of this, where I expect to see, you know, MCPs really oriented around the cloud desktop app for a reason, right? Like, I think a lot of these tools are going to run on your computer because it makes... I think it's breaking out. People are putting it on a server. Oh, really? Okay. Well, sweet. We'll see. We'll see that. I was surprised, though, wasn't I? I think that the browser company, too, with Dia Browser, it runs on your machine. You know, it's going to be...swyx [00:30:50]: What is it?Paul [00:30:51]: So, Dia Browser, as far as I understand... I used to use Arc. Yeah. I haven't used Arc. But I'm a big fan of the browser company. I think they're doing a lot of cool stuff in consumer. As far as I understand, it's a browser where you have a sidebar where you can, like, chat with it and it can control the local browser on your machine. So, if you imagine, like, what a consumer web agent is, which it lives alongside your browser, I think Google Chrome has Project Marina, I think. I almost call it Project Marinara for some reason. I don't know why. It's...swyx [00:31:17]: No, I think it's someone really likes the Waterworld. Oh, I see. The classic Kevin Costner. Yeah.Paul [00:31:22]: Okay. Project Marinara is a similar thing to the Dia Browser, in my mind, as far as I understand it. You have a browser that has an AI interface that will take over your mouse and keyboard and control the browser for you. Great for consumer use cases. But if you're building applications that rely on a browser and it's more part of a greater, like, AI app experience, you probably need something that's more like infrastructure, not a consumer app.swyx [00:31:44]: Just because I have explored a little bit in this area, do people want branching? So, I have the state. Of whatever my browser's in. And then I want, like, 100 clones of this state. Do people do that? Or...Paul [00:31:56]: People don't do it currently. Yeah. But it's definitely something we're thinking about. I think the idea of forking a browser is really cool. Technically, kind of hard. We're starting to see this in code execution, where people are, like, forking some, like, code execution, like, processes or forking some tool calls or branching tool calls. Haven't seen it at the browser level yet. But it makes sense. Like, if an AI agent is, like, using a website and it's not sure what path it wants to take to crawl this website. To find the information it's looking for. It would make sense for it to explore both paths in parallel. And that'd be a very, like... A road not taken. Yeah. And hopefully find the right answer. And then say, okay, this was actually the right one. And memorize that. And go there in the future. On the roadmap. For sure. Don't make my roadmap, please. You know?Alessio [00:32:37]: How do you actually do that? Yeah. How do you fork? I feel like the browser is so stateful for so many things.swyx [00:32:42]: Serialize the state. Restore the state. I don't know.Paul [00:32:44]: So, it's one of the reasons why we haven't done it yet. It's hard. You know? Like, to truly fork, it's actually quite difficult. The naive way is to open the same page in a new tab and then, like, hope that it's at the same thing. But if you have a form halfway filled, you may have to, like, take the whole, you know, container. Pause it. All the memory. Duplicate it. Restart it from there. It could be very slow. So, we haven't found a thing. Like, the easy thing to fork is just, like, copy the page object. You know? But I think there needs to be something a little bit more robust there. Yeah.swyx [00:33:12]: So, MorphLabs has this infinite branch thing. Like, wrote a custom fork of Linux or something that let them save the system state and clone it. MorphLabs, hit me up. I'll be a customer. Yeah. That's the only. I think that's the only way to do it. Yeah. Like, unless Chrome has some special API for you. Yeah.Paul [00:33:29]: There's probably something we'll reverse engineer one day. I don't know. Yeah.Alessio [00:33:32]: Let's talk about StageHand, the AI web browsing framework. You have three core components, Observe, Extract, and Act. Pretty clean landing page. What was the idea behind making a framework? Yeah.Stagehand: AI web browsing frameworkPaul [00:33:43]: So, there's three frameworks that are very popular or already exist, right? Puppeteer, Playwright, Selenium. Those are for building hard-coded scripts to control websites. And as soon as I started to play with LLMs plus browsing, I caught myself, you know, code-genning Playwright code to control a website. I would, like, take the DOM. I'd pass it to an LLM. I'd say, can you generate the Playwright code to click the appropriate button here? And it would do that. And I was like, this really should be part of the frameworks themselves. And I became really obsessed with SDKs that take natural language as part of, like, the API input. And that's what StageHand is. StageHand exposes three APIs, and it's a super set of Playwright. So, if you go to a page, you may want to take an action, click on the button, fill in the form, etc. That's what the act command is for. You may want to extract some data. This one takes a natural language, like, extract the winner of the Super Bowl from this page. You can give it a Zod schema, so it returns a structured output. And then maybe you're building an API. You can do an agent loop, and you want to kind of see what actions are possible on this page before taking one. You can do observe. So, you can observe the actions on the page, and it will generate a list of actions. You can guide it, like, give me actions on this page related to buying an item. And you can, like, buy it now, add to cart, view shipping options, and pass that to an LLM, an agent loop, to say, what's the appropriate action given this high-level goal? So, StageHand isn't a web agent. It's a framework for building web agents. And we think that agent loops are actually pretty close to the application layer because every application probably has different goals or different ways it wants to take steps. I don't think I've seen a generic. Maybe you guys are the experts here. I haven't seen, like, a really good AI agent framework here. Everyone kind of has their own special sauce, right? I see a lot of developers building their own agent loops, and they're using tools. And I view StageHand as the browser tool. So, we expose act, extract, observe. Your agent can call these tools. And from that, you don't have to worry about it. You don't have to worry about generating playwright code performantly. You don't have to worry about running it. You can kind of just integrate these three tool calls into your agent loop and reliably automate the web.swyx [00:35:48]: A special shout-out to Anirudh, who I met at your dinner, who I think listens to the pod. Yeah. Hey, Anirudh.Paul [00:35:54]: Anirudh's a man. He's a StageHand guy.swyx [00:35:56]: I mean, the interesting thing about each of these APIs is they're kind of each startup. Like, specifically extract, you know, Firecrawler is extract. There's, like, Expand AI. There's a whole bunch of, like, extract companies. They just focus on extract. I'm curious. Like, I feel like you guys are going to collide at some point. Like, right now, it's friendly. Everyone's in a blue ocean. At some point, it's going to be valuable enough that there's some turf battle here. I don't think you have a dog in a fight. I think you can mock extract to use an external service if they're better at it than you. But it's just an observation that, like, in the same way that I see each option, each checkbox in the side of custom GBTs becoming a startup or each box in the Karpathy chart being a startup. Like, this is also becoming a thing. Yeah.Paul [00:36:41]: I mean, like, so the way StageHand works is that it's MIT-licensed, completely open source. You bring your own API key to your LLM of choice. You could choose your LLM. We don't make any money off of the extract or really. We only really make money if you choose to run it with our browser. You don't have to. You can actually use your own browser, a local browser. You know, StageHand is completely open source for that reason. And, yeah, like, I think if you're building really complex web scraping workflows, I don't know if StageHand is the tool for you. I think it's really more if you're building an AI agent that needs a few general tools or if it's doing a lot of, like, web automation-intensive work. But if you're building a scraping company, StageHand is not your thing. You probably want something that's going to, like, get HTML content, you know, convert that to Markdown, query it. That's not what StageHand does. StageHand is more about reliability. I think we focus a lot on reliability and less so on cost optimization and speed at this point.swyx [00:37:33]: I actually feel like StageHand, so the way that StageHand works, it's like, you know, page.act, click on the quick start. Yeah. It's kind of the integration test for the code that you would have to write anyway, like the Puppeteer code that you have to write anyway. And when the page structure changes, because it always does, then this is still the test. This is still the test that I would have to write. Yeah. So it's kind of like a testing framework that doesn't need implementation detail.Paul [00:37:56]: Well, yeah. I mean, Puppeteer, Playwright, and Slenderman were all designed as testing frameworks, right? Yeah. And now people are, like, hacking them together to automate the web. I would say, and, like, maybe this is, like, me being too specific. But, like, when I write tests, if the page structure changes. Without me knowing, I want that test to fail. So I don't know if, like, AI, like, regenerating that. Like, people are using StageHand for testing. But it's more for, like, usability testing, not, like, testing of, like, does the front end, like, has it changed or not. Okay. But generally where we've seen people, like, really, like, take off is, like, if they're using, you know, something. If they want to build a feature in their application that's kind of like Operator or Deep Research, they're using StageHand to kind of power that tool calling in their own agent loop. Okay. Cool.swyx [00:38:37]: So let's go into Operator, the first big agent launch of the year from OpenAI. Seems like they have a whole bunch scheduled. You were on break and your phone blew up. What's your just general view of computer use agents is what they're calling it. The overall category before we go into Open Operator, just the overall promise of Operator. I will observe that I tried it once. It was okay. And I never tried it again.OpenAI's Operator and computer use agentsPaul [00:38:58]: That tracks with my experience, too. Like, I'm a huge fan of the OpenAI team. Like, I think that I do not view Operator as the company. I'm not a company killer for browser base at all. I think it actually shows people what's possible. I think, like, computer use models make a lot of sense. And I'm actually most excited about computer use models is, like, their ability to, like, really take screenshots and reasoning and output steps. I think that using mouse click or mouse coordinates, I've seen that proved to be less reliable than I would like. And I just wonder if that's the right form factor. What we've done with our framework is anchor it to the DOM itself, anchor it to the actual item. So, like, if it's clicking on something, it's clicking on that thing, you know? Like, it's more accurate. No matter where it is. Yeah, exactly. Because it really ties in nicely. And it can handle, like, the whole viewport in one go, whereas, like, Operator can only handle what it sees. Can you hover? Is hovering a thing that you can do? I don't know if we expose it as a tool directly, but I'm sure there's, like, an API for hovering. Like, move mouse to this position. Yeah, yeah, yeah. I think you can trigger hover, like, via, like, the JavaScript on the DOM itself. But, no, I think, like, when we saw computer use, everyone's eyes lit up because they realized, like, wow, like, AI is going to actually automate work for people. And I think seeing that kind of happen from both of the labs, and I'm sure we're going to see more labs launch computer use models, I'm excited to see all the stuff that people build with it. I think that I'd love to see computer use power, like, controlling a browser on browser base. And I think, like, Open Operator, which was, like, our open source version of OpenAI's Operator, was our first take on, like, how can we integrate these models into browser base? And we handle the infrastructure and let the labs do the models. I don't have a sense that Operator will be released as an API. I don't know. Maybe it will. I'm curious to see how well that works because I think it's going to be really hard for a company like OpenAI to do things like support CAPTCHA solving or, like, have proxies. Like, I think it's hard for them structurally. Imagine this New York Times headline, OpenAI CAPTCHA solving. Like, that would be a pretty bad headline, this New York Times headline. Browser base solves CAPTCHAs. No one cares. No one cares. And, like, our investors are bored. Like, we're all okay with this, you know? We're building this company knowing that the CAPTCHA solving is short-lived until we figure out how to authenticate good bots. I think it's really hard for a company like OpenAI, who has this brand that's so, so good, to balance with, like, the icky parts of web automation, which it can be kind of complex to solve. I'm sure OpenAI knows who to call whenever they need you. Yeah, right. I'm sure they'll have a great partnership.Alessio [00:41:23]: And is Open Operator just, like, a marketing thing for you? Like, how do you think about resource allocation? So, you can spin this up very quickly. And now there's all this, like, open deep research, just open all these things that people are building. We started it, you know. You're the original Open. We're the original Open operator, you know? Is it just, hey, look, this is a demo, but, like, we'll help you build out an actual product for yourself? Like, are you interested in going more of a product route? That's kind of the OpenAI way, right? They started as a model provider and then…Paul [00:41:53]: Yeah, we're not interested in going the product route yet. I view Open Operator as a model provider. It's a reference project, you know? Let's show people how to build these things using the infrastructure and models that are out there. And that's what it is. It's, like, Open Operator is very simple. It's an agent loop. It says, like, take a high-level goal, break it down into steps, use tool calling to accomplish those steps. It takes screenshots and feeds those screenshots into an LLM with the step to generate the right action. It uses stagehand under the hood to actually execute this action. It doesn't use a computer use model. And it, like, has a nice interface using the live view that we talked about, the iframe, to embed that into an application. So I felt like people on launch day wanted to figure out how to build their own version of this. And we turned that around really quickly to show them. And I hope we do that with other things like deep research. We don't have a deep research launch yet. I think David from AOMNI actually has an amazing open deep research that he launched. It has, like, 10K GitHub stars now. So he's crushing that. But I think if people want to build these features natively into their application, they need good reference projects. And I think Open Operator is a good example of that.swyx [00:42:52]: I don't know. Actually, I'm actually pretty bullish on API-driven operator. Because that's the only way that you can sort of, like, once it's reliable enough, obviously. And now we're nowhere near. But, like, give it five years. It'll happen, you know. And then you can sort of spin this up and browsers are working in the background and you don't necessarily have to know. And it just is booking restaurants for you, whatever. I can definitely see that future happening. I had this on the landing page here. This might be a slightly out of order. But, you know, you have, like, sort of three use cases for browser base. Open Operator. Or this is the operator sort of use case. It's kind of like the workflow automation use case. And it completes with UiPath in the sort of RPA category. Would you agree with that? Yeah, I would agree with that. And then there's Agents we talked about already. And web scraping, which I imagine would be the bulk of your workload right now, right?Paul [00:43:40]: No, not at all. I'd say actually, like, the majority is browser automation. We're kind of expensive for web scraping. Like, I think that if you're building a web scraping product, if you need to do occasional web scraping or you have to do web scraping that works every single time, you want to use browser automation. Yeah. You want to use browser-based. But if you're building web scraping workflows, what you should do is have a waterfall. You should have the first request is a curl to the website. See if you can get it without even using a browser. And then the second request may be, like, a scraping-specific API. There's, like, a thousand scraping APIs out there that you can use to try and get data. Scraping B. Scraping B is a great example, right? Yeah. And then, like, if those two don't work, bring out the heavy hitter. Like, browser-based will 100% work, right? It will load the page in a real browser, hydrate it. I see.swyx [00:44:21]: Because a lot of people don't render to JS.swyx [00:44:25]: Yeah, exactly.Paul [00:44:26]: So, I mean, the three big use cases, right? Like, you know, automation, web data collection, and then, you know, if you're building anything agentic that needs, like, a browser tool, you want to use browser-based.Alessio [00:44:35]: Is there any use case that, like, you were super surprised by that people might not even think about? Oh, yeah. Or is it, yeah, anything that you can share? The long tail is crazy. Yeah.Surprising use cases of BrowserbasePaul [00:44:44]: One of the case studies on our website that I think is the most interesting is this company called Benny. So, the way that it works is if you're on food stamps in the United States, you can actually get rebates if you buy certain things. Yeah. You buy some vegetables. You submit your receipt to the government. They'll give you a little rebate back. Say, hey, thanks for buying vegetables. It's good for you. That process of submitting that receipt is very painful. And the way Benny works is you use their app to take a photo of your receipt, and then Benny will go submit that receipt for you and then deposit the money into your account. That's actually using no AI at all. It's all, like, hard-coded scripts. They maintain the scripts. They've been doing a great job. And they build this amazing consumer app. But it's an example of, like, all these, like, tedious workflows that people have to do to kind of go about their business. And they're doing it for the sake of their day-to-day lives. And I had never known about, like, food stamp rebates or the complex forms you have to do to fill them. But the world is powered by millions and millions of tedious forms, visas. You know, Emirate Lighthouse is a customer, right? You know, they do the O1 visa. Millions and millions of forms are taking away humans' time. And I hope that Browserbase can help power software that automates away the web forms that we don't need anymore. Yeah.swyx [00:45:49]: I mean, I'm very supportive of that. I mean, forms. I do think, like, government itself is a big part of it. I think the government itself should embrace AI more to do more sort of human-friendly form filling. Mm-hmm. But I'm not optimistic. I'm not holding my breath. Yeah. We'll see. Okay. I think I'm about to zoom out. I have a little brief thing on computer use, and then we can talk about founder stuff, which is, I tend to think of developer tooling markets in impossible triangles, where everyone starts in a niche, and then they start to branch out. So I already hinted at a little bit of this, right? We mentioned more. We mentioned E2B. We mentioned Firecrawl. And then there's Browserbase. So there's, like, all this stuff of, like, have serverless virtual computer that you give to an agent and let them do stuff with it. And there's various ways of connecting it to the internet. You can just connect to a search API, like SERP API, whatever other, like, EXA is another one. That's what you're searching. You can also have a JSON markdown extractor, which is Firecrawl. Or you can have a virtual browser like Browserbase, or you can have a virtual machine like Morph. And then there's also maybe, like, a virtual sort of code environment, like Code Interpreter. So, like, there's just, like, a bunch of different ways to tackle the problem of give a computer to an agent. And I'm just kind of wondering if you see, like, everyone's just, like, happily coexisting in their respective niches. And as a developer, I just go and pick, like, a shopping basket of one of each. Or do you think that you eventually, people will collide?Future of browser automation and market competitionPaul [00:47:18]: I think that currently it's not a zero-sum market. Like, I think we're talking about... I think we're talking about all of knowledge work that people do that can be automated online. All of these, like, trillions of hours that happen online where people are working. And I think that there's so much software to be built that, like, I tend not to think about how these companies will collide. I just try to solve the problem as best as I can and make this specific piece of infrastructure, which I think is an important primitive, the best I possibly can. And yeah. I think there's players that are actually going to like it. I think there's players that are going to launch, like, over-the-top, you know, platforms, like agent platforms that have all these tools built in, right? Like, who's building the rippling for agent tools that has the search tool, the browser tool, the operating system tool, right? There are some. There are some. There are some, right? And I think in the end, what I have seen as my time as a developer, and I look at all the favorite tools that I have, is that, like, for tools and primitives with sufficient levels of complexity, you need to have a solution that's really bespoke to that primitive, you know? And I am sufficiently convinced that the browser is complex enough to deserve a primitive. Obviously, I have to. I'm the founder of BrowserBase, right? I'm talking my book. But, like, I think maybe I can give you one spicy take against, like, maybe just whole OS running. I think that when I look at computer use when it first came out, I saw that the majority of use cases for computer use were controlling a browser. And do we really need to run an entire operating system just to control a browser? I don't think so. I don't think that's necessary. You know, BrowserBase can run browsers for way cheaper than you can if you're running a full-fledged OS with a GUI, you know, operating system. And I think that's just an advantage of the browser. It is, like, browsers are little OSs, and you can run them very efficiently if you orchestrate it well. And I think that allows us to offer 90% of the, you know, functionality in the platform needed at 10% of the cost of running a full OS. Yeah.Open Operator: Browserbase's Open-Source Alternativeswyx [00:49:16]: I definitely see the logic in that. There's a Mark Andreessen quote. I don't know if you know this one. Where he basically observed that the browser is turning the operating system into a poorly debugged set of device drivers, because most of the apps are moved from the OS to the browser. So you can just run browsers.Paul [00:49:31]: There's a place for OSs, too. Like, I think that there are some applications that only run on Windows operating systems. And Eric from pig.dev in this upcoming YC batch, or last YC batch, like, he's building all run tons of Windows operating systems for you to control with your agent. And like, there's some legacy EHR systems that only run on Internet-controlled systems. Yeah.Paul [00:49:54]: I think that's it. I think, like, there are use cases for specific operating systems for specific legacy software. And like, I'm excited to see what he does with that. I just wanted to give a shout out to the pig.dev website.swyx [00:50:06]: The pigs jump when you click on them. Yeah. That's great.Paul [00:50:08]: Eric, he's the former co-founder of banana.dev, too.swyx [00:50:11]: Oh, that Eric. Yeah. That Eric. Okay. Well, he abandoned bananas for pigs. I hope he doesn't start going around with pigs now.Alessio [00:50:18]: Like he was going around with bananas. A little toy pig. Yeah. Yeah. I love that. What else are we missing? I think we covered a lot of, like, the browser-based product history, but. What do you wish people asked you? Yeah.Paul [00:50:29]: I wish people asked me more about, like, what will the future of software look like? Because I think that's really where I've spent a lot of time about why do browser-based. Like, for me, starting a company is like a means of last resort. Like, you shouldn't start a company unless you absolutely have to. And I remain convinced that the future of software is software that you're going to click a button and it's going to do stuff on your behalf. Right now, software. You click a button and it maybe, like, calls it back an API and, like, computes some numbers. It, like, modifies some text, whatever. But the future of software is software using software. So, I may log into my accounting website for my business, click a button, and it's going to go load up my Gmail, search my emails, find the thing, upload the receipt, and then comment it for me. Right? And it may use it using APIs, maybe a browser. I don't know. I think it's a little bit of both. But that's completely different from how we've built software so far. And that's. I think that future of software has different infrastructure requirements. It's going to require different UIs. It's going to require different pieces of infrastructure. I think the browser infrastructure is one piece that fits into that, along with all the other categories you mentioned. So, I think that it's going to require developers to think differently about how they've built software for, you know

Read the article here: https://journals.sagepub.com/doi/full/10.1177/30494826241296390

In this final episode of the four-part series on hepatocellular carcinoma (HCC), hosted by the Oncology Brothers, Drs Rohit and Rahul Gosain, the discussion focuses on the evolving role of immunotherapy (IO) in intermediate HCC. The episode explores multimodal approaches that combine IO and IO-based therapies with loco-regional treatments and highlights the essential role of a multidisciplinary care team. Drs Nina Sanford (radiation oncologist), Mark Yarchoan (medical oncologist), and Ed Kim (interventional radiologist) join the Oncology Brothers to share their insights on: • Current treatment options for intermediate HCC, addressing its heterogeneity and standard treatment pathways • Latest clinical trial data (EMERALD-1, LEAP-012) on combining IO with loco-regional therapies, and the clinical implications • The importance of effective collaboration within the multidisciplinary team for delivering optimal patient care • Combining IO with loco-regional therapy and future perspectives in the field Clinical takeaways • IO and IO-based treatments are moving earlier in the treatment paradigm for patients with intermediate HCC. Earlier integration of these therapies aims to achieve improved systemic control, allowing loco-regional therapy to target oligoprogression, residual lesions or reduce tumour burden • Emerging data supports combining systemic and loco-regional therapies for patients with intermediate HCC. EMERALD-1 and LEAP-012 show promising PFS data using IO-based combination regimens like durvalumab + bevacizumab or pembrolizumab + lenvatinib alongside TACE. Long-term OS data are awaited • Effective communication and coordinated care among specialists, such as medical oncologists, radiation oncologists, hepatologists, and interventional radiologists, are essential to developing optimal treatment strategies for patients with intermediate HCC Follow us on social media: •⁠ ⁠X/Twitter: https://twitter.com/oncbrothers •⁠ ⁠Instagram: https://www.instagram.com/oncbrothers •⁠ YouTube: https://www.youtube.com/channel/UCjfxKlVho5xWH5ltufj4F4A/ Subscribe to our channel for more insights on oncology treatments and patient care!

In this episode I interviewed Wee Kee, the co-founder of Virtuals. From here and there in the conversation you can tell that Virtuals team is a pragmatic and perseverant one. To build a successful product in this fast moving industry, there's no fast track but stay on the table and keep grinding. Shownotes:You guys have been dominating mindshare amongst all the crypto native folks lately. So first of all congratulations! But still would love to hear how you introduce yourself to our audience. There were quite many platforms that empower developers to launch agents. What key factors do you think led you to where you guys are today?Do you think the diversity of the agents came from the no-code requirement?What are some of the agents that grew from Virtuals that left people an impression? Also about Multimodality. What are the best performing categories for agents developed with GAME?How do you decide whether you're going to co-marketing with one of the ecosystem agents?I read on Messari that you guys have integrated Farcaster, but I haven't seen a lot of news that you guys are pushing on that front. Any plans? I was particularly interested to hear your answer on this one, because looks like with a web3 native social network, an agent developed under GAME framework will be capable of doing a lot more

Erum and Karl have an incredible chat with Nick Edwards, the innovative mind behind Potato AI. Nick goes into the inspiration behind the platform's unique name, rooted in childhood curiosity and scientific wonder, and shares his journey from neuroscience research to building a groundbreaking tool aimed at accelerating scientific discovery. He explains how Potato tackles the overwhelming deluge of scientific literature and the reproducibility crisis using AI to structure and analyze data in transformative ways. With stories from his own career, a stint in consulting, and running his podcast "Once a Scientist," Nick discusses the future of AI, science, and collaboration. This episode has amazing insights into how technology is reshaping the very fabric of research. Grow Everything brings the bioeconomy to life. Hosts Karl Schmieder and Erum Azeez Khan share stories and interview the leaders and influencers changing the world by growing everything. Biology is the oldest technology. And it can be engineered. What are we growing? Learn more at ⁠⁠⁠⁠www.messaginglab.com/groweverything⁠⁠⁠⁠ Chapters: 00:00:00 - Potatoes, Balloons, and a DIY Lab Setup 00:04:56 - Reproducibility Crisis: 70% of Research in Question 00:06:40 - Meet Nick Edwards: The Potato Visionary 00:10:00 - Organizing Scientific Chaos: How AI Helps 00:14:47 - AI Research Assistant vs. AI Scientist: The Journey 00:20:09 - From Neuroscience to Entrepreneurship: A Scientist's Leap 00:25:00 - Biopharma Meets Bioindustrials: Bridging Two Worlds 00:29:27 - Liquid Handlers and Shared Protocols: Optimizing Biotech 00:35:13 - Multimodality and the AI Scientist Dream 00:39:04 - Citizen Scientists and Accessible AI Tools 00:45:00 - The Next Scientific Revolution: AI and Decentralized Science Episode Links: Potato AI Once A Scientist Podcast Merck Digital Science Studio Wiley Elsevier Ginkgo Bioworks Topics Covered: Research, AI research assistant, AI Scientist, Biotech, Lab Automation, Reproducibility Have a question or comment? Message us here: Text or Call (804) 505-5553 ⁠⁠⁠⁠Instagram⁠⁠⁠⁠  / ⁠⁠⁠⁠Twitter⁠⁠⁠⁠ / ⁠⁠⁠⁠LinkedIn⁠⁠⁠⁠ / ⁠⁠⁠⁠Youtube⁠⁠⁠⁠ / ⁠⁠⁠⁠Grow Everything⁠⁠⁠⁠ Email: groweverything@messaginglab.com Music by: Nihilore Production by: Amplafy Media

Nathan welcomes back computational biochemist Amelie Schreiber for a fascinating update on AI's revolutionary impact in biology. In this episode of The Cognitive Revolution, we explore recent breakthroughs including AlphaFold3, ESM3, and new diffusion models transforming protein engineering and drug discovery. Join us for an insightful discussion about how AI is reshaping our understanding of molecular biology and making complex protein engineering tasks more accessible than ever before. Help shape our show by taking our quick listener survey at https://bit.ly/TurpentinePulse SPONSORS: Shopify: Shopify is the world's leading e-commerce platform, offering a market-leading checkout system and exclusive AI apps like Quikly. Nobody does selling better than Shopify. Get a $1 per month trial at https://shopify.com/cognitive SelectQuote: Finding the right life insurance shouldn't be another task you put off. SelectQuote compares top-rated policies to get you the best coverage at the right price. Even in our AI-driven world, protecting your family's future remains essential. Get your personalized quote at https://selectquote.com/cognitive Oracle Cloud Infrastructure (OCI): Oracle's next-generation cloud platform delivers blazing-fast AI and ML performance with 50% less for compute and 80% less for outbound networking compared to other cloud providers13. OCI powers industry leaders with secure infrastructure and application development capabilities. New U.S. customers can get their cloud bill cut in half by switching to OCI before December 31, 2024 at https://oracle.com/cognitive Weights & Biases RAG++: Advanced training for building production-ready RAG applications. Learn from experts to overcome LLM challenges, evaluate systematically, and integrate advanced features. Includes free Cohere credits. Visit https://wandb.me/cr to start the RAG++ course today. CHAPTERS: (00:00:00) Teaser (00:00:46) About the Episode (00:04:30) AI for Biology (00:07:14) David Baker's Impact (00:11:49) AlphaFold 3 & ESM3 (00:16:40) Protein Interaction Prediction (Part 1) (00:16:44) Sponsors: Shopify | SelectQuote (00:19:18) Protein Interaction Prediction (Part 2) (00:31:12) MSAs & Embeddings (Part 1) (00:32:32) Sponsors: Oracle Cloud Infrastructure (OCI) | Weights & Biases RAG++ (00:34:49) MSAs & Embeddings (Part 2) (00:35:57) Beyond Structure Prediction (00:51:13) Dynamics vs. Statics (00:57:24) In-Painting & Use Cases (00:59:48) Workflow & Platforms (01:06:45) Design Process & Success Rates (01:13:23) Ambition & Task Definition (01:19:25) New Models: PepFlow & GeoAB (01:28:23) Flow Matching vs. Diffusion (01:30:42) ESM3 & Multimodality (01:37:10) Summary & Future Directions (01:45:34) Outro SOCIAL LINKS: Website: https://www.cognitiverevolution.ai Twitter (Podcast): https://x.com/cogrev_podcast Twitter (Nathan): https://x.com/labenz LinkedIn: https://www.linkedin.com/in/nathanlabenz/ Youtube: https://www.youtube.com/@CognitiveRevolutionPodcast Apple: https://podcasts.apple.com/de/podcast/the-cognitive-revolution-ai-builders-researchers-and/id1669813431

Hey folks, Alex here, writing this from the beautiful Vancouver BC, Canada. I'm here for NeurIPS 2024, the biggest ML conferences of the year, and let me tell you, this was one hell of a week to not be glued to the screen. After last week banger week, with OpenAI kicking off their 12 days of releases, with releasing o1 full and pro mode during ThursdAI, things went parabolic. It seems that all the AI labs decided to just dump EVERYTHING they have before the holidays?

Nathan interviews Google product managers Shrestha Basu Mallick and Logan Kilpatrick about the Gemini API and AI Studio. They discuss Google's new grounding feature, allowing Gemini models to access real-time web information via Google search. The conversation explores Gemini's rapid growth, its position in the AI landscape, and Google's competitive strategy. Nathan shares insights from integrating Gemini into his own application and ponders the future of large language model capabilities across providers. Tune in for an in-depth look at Google's AI API product strategy and the latest Gemini features. Be notified early when Turpentine's drops new publication: https://www.turpentine.co/exclusiveaccess SPONSORS: Weights & Biases RAG++: Advanced training for building production-ready RAG applications. Learn from experts to overcome LLM challenges, evaluate systematically, and integrate advanced features. Includes free Cohere credits. Visit https://wandb.me/cr to start the RAG++ course today. Shopify: Shopify is the world's leading e-commerce platform, offering a market-leading checkout system and exclusive AI apps like Quikly. Nobody does selling better than Shopify. Get a $1 per month trial at https://shopify.com/cognitive Notion: Notion offers powerful workflow and automation templates, perfect for streamlining processes and laying the groundwork for AI-driven automation. With Notion AI, you can search across thousands of documents from various platforms, generating highly relevant analysis and content tailored just for you - try it for free at https://notion.com/cognitiverevolution LMNT: LMNT is a zero-sugar electrolyte drink mix that's redefining hydration and performance. Ideal for those who fast or anyone looking to optimize their electrolyte intake. Support the show and get a free sample pack with any purchase at https://drinklmnt.com/tcr CHAPTERS: (00:00:00) About the Show (00:00:53) Sponsors: Weights & Biases RAG++ (00:01:28) About the Episode (00:04:15) Gemini API Growth (00:05:26) Intro to AI Studio (00:07:35) Vertex vs. AI Studio (00:09:33) Developer Adoption (00:14:23) Gemini Use Cases (Part 1) (00:17:41) Sponsors: Shopify | Notion (00:20:01) Gemini Use Cases (Part 2) (00:23:08) Multimodality & Flash (00:26:29) Free Tier & Costs (00:31:43) Inference Costs (00:32:55) Fine-tuning & Vision (00:36:59) Sponsors: LMNT (00:38:04) Search Grounding (00:44:42) Grounding Sources (00:46:58) Competitive Landscape (00:50:36) Design Decisions (00:54:54) Outro SOCIAL LINKS: Website: https://www.cognitiverevolution.ai Twitter (Podcast): https://x.com/cogrev_podcast Twitter (Nathan): https://x.com/labenz LinkedIn: https://www.linkedin.com/in/nathanlabenz/ Youtube: https://www.youtube.com/@CognitiveRevolutionPodcast Apple: https://podcasts.apple.com/de/podcast/the-cognitive-revolution-ai-builders-researchers-and/id1669813431 Spotify: https://open.spotify.com/show/6yHyok3M3BjqzR0VB5MSyk

In this podcast, we discuss the The Society of Thoracic Surgeons/American Society for Radiation Oncology Updated Clinical Practice Guidelines on Multimodality Therapy for Locally Advanced Cancer of the Esophagus or Gastroesophageal Junction. Joining in the discussion are Dr. Stephanie Worrell, Associate Professor and Thoracic Section Chief in the Division of Cardiothoracic Surgery at the University of Arizona College of Medicine, and Dr. Karyn Goodman, Professor and Vice Chair for Research and Quality at the Icahn School of Medicine at Mount Sinai, and Associate Director for Clinical Research at The Tisch Cancer Institute, who served as chair and co-chair of the guideline panel, respectively. Together, we cover important updates and recommendations that incorporate surgical aspects into the multi-disciplinary management of this disease along with practical considerations for everyday practice. Additionally, we discuss in depth the recently presented ESOPEC trial presented at the 2024 ASCO annual meeting and how it has impacted the standard of care for esophageal cancers.

In this podcast, we discuss the The Society of Thoracic Surgeons/American Society for Radiation Oncology Updated Clinical Practice Guidelines on Multimodality Therapy for Locally Advanced Cancer of the Esophagus or Gastroesophageal Junction. Joining in the discussion are Dr. Stephanie Worrell, Associate Professor and Thoracic Section Chief in the Division of Cardiothoracic Surgery at the University of Arizona College of Medicine, and Dr. Karyn Goodman, Professor and Vice Chair for Research and Quality at the Icahn School of Medicine at Mount Sinai, and Associate Director for Clinical Research at The Tisch Cancer Institute, who served as chair and co-chair of the guideline panel, respectively. Together, we cover important updates and recommendations that incorporate surgical aspects into the multi-disciplinary management of this disease along with practical considerations for everyday practice. Additionally, we discuss in depth the recently presented ESOPEC trial presented at the 2024 ASCO annual meeting and how it has impacted the standard of care for esophageal cancers.

In this podcast, we discuss the The Society of Thoracic Surgeons/American Society for Radiation Oncology Updated Clinical Practice Guidelines on Multimodality Therapy for Locally Advanced Cancer of the Esophagus or Gastroesophageal Junction. Joining in the discussion are Dr. Stephanie Worrell, Associate Professor and Thoracic Section Chief in the Division of Cardiothoracic Surgery at the University of Arizona College of Medicine, and Dr. Karyn Goodman, Professor and Vice Chair for Research and Quality at the Icahn School of Medicine at Mount Sinai, and Associate Director for Clinical Research at The Tisch Cancer Institute, who served as chair and co-chair of the guideline panel, respectively. Together, we cover important updates and recommendations that incorporate surgical aspects into the multi-disciplinary management of this disease along with practical considerations for everyday practice. Additionally, we discuss in depth the recently presented ESOPEC trial presented at the 2024 ASCO annual meeting and how it has impacted the standard of care for esophageal cancers.

In this podcast, we discuss the The Society of Thoracic Surgeons/American Society for Radiation Oncology Updated Clinical Practice Guidelines on Multimodality Therapy for Locally Advanced Cancer of the Esophagus or Gastroesophageal Junction. Joining in the discussion are Dr. Stephanie Worrell, Associate Professor and Thoracic Section Chief in the Division of Cardiothoracic Surgery at the University of Arizona College of Medicine, and Dr. Karyn Goodman, Professor and Vice Chair for Research and Quality at the Icahn School of Medicine at Mount Sinai, and Associate Director for Clinical Research at The Tisch Cancer Institute, who served as chair and co-chair of the guideline panel, respectively. Together, we cover important updates and recommendations that incorporate surgical aspects into the multi-disciplinary management of this disease along with practical considerations for everyday practice. Additionally, we discuss in depth the recently presented ESOPEC trial presented at the 2024 ASCO annual meeting and how it has impacted the standard of care for esophageal cancers.

CardioNerds (Dr. Dan Ambinder and Dr. Rick Ferraro) join Dr. Mansi Oberoi and Dr. Mohan Gudiwada from the University of Nebraska Medical Center discuss a case of constrictive pericarditis. Expert commentary is provided by Dr. Adam Burdorf, who serves as the Program Director for the Cardiovascular Medicine Fellowship at the University of Nebraska Medical Center. The case discussed involves a 76-year-old woman with a history of monoclonal gammopathy of undetermined significance, chronic obstructive pulmonary disease, type 2 diabetes mellitus, and squamous cell carcinoma was admitted to the hospital for worsening shortness of breath, swelling in lower extremities, hyponatremia, and urinary tract infection. CT chest to evaluate for pulmonary embolism showed incidental pericardial calcifications; the heart failure team was consulted for the management of her decompensated heart failure. Echo images were nondiagnostic. Subsequent invasive hemodynamic monitoring showed elevated right and left-sided filling pressures, diastolic equalization of LV and RV pressures, and positive RV square root sign with ventricular interdependence. Cardiac MRI showed septal flattening on deep inspiration and septal bounce, suggestive of interventricular dependence. After a heart team discussion and with shared-decision making the patient opted for medical management owing to her comorbidities and frailty. Enjoy this 2024 JACC State-of-the-Art Review to learn more about pericardial diseases and best practices for pericardiectomy (Al-Kazac et al., JACC 2024) US Cardiology Review is now the official journal of CardioNerds! Submit your manuscript here. CardioNerds Case Reports PageCardioNerds Episode PageCardioNerds AcademyCardionerds Healy Honor Roll CardioNerds Journal ClubSubscribe to The Heartbeat Newsletter!Check out CardioNerds SWAG!Become a CardioNerds Patron! Case Media - Constrictive Pericarditis Echo: Left Ventricular ejection fraction = 55-60%. Unclear septal motion in the setting of atrial fibrillation MRI: Diastolic septal flattening with deep inspiration as well as a septal bounce suggestive of interventricular dependence and constrictive physiology  References Garcia, M. Constrictive Pericarditis Versus Restrictive Cardiomyopathy. Journal of the American College of Cardiology, vol. 67, no. 17, 2016, pp. 2061–2076. Pathophysiology and Diagnosis of Constrictive Pericarditis. American College of Cardiology, 2017. Geske, J., Anavekar, N., Nishimura, R., et al. Differentiation of Constriction and Restriction: Complex Cardiovascular Hemodynamics. Journal of the American College of Cardiology, vol. 68, no. 21, 2016, pp. 2329–2347. Constrictive Pericarditis. ScienceDirect. Constrictive Pericarditis. Journal of the American College of Cardiology, vol. 83, no. 12, 2024, pp. 1500-1512.

Descript CEO and founder Andrew Mason joins Lightspeed Partner and Host Michael Mignano on the podcast to talk about the future of content creation with AI tools. Michael and Andrew talk about the evolution of Descript as an AI product designed for podcast and video creators, navigating a world of synthetic content, and Descript's new features including Descript Rooms and Underlord. Andrew talks about raising a $50 Million Series C led by OpenAI Startup Fund and how seeing an early version of ChatGPT inspired confidence in Descript's foundational vision and goal to simplify media production.   Episode Chapters (00:00) Introduction(00:09) Introducing Descript Rooms(01:10) Descript's Versatility with Media Creation(04:22) Social Clips and Longform Content(07:15) Craft and Control in AI in Content Creation(13:16) Descript AI Tools, OpenAI, and ChatGPT(17:30) Multimodality and Improving Quality(26:17) Trust and Adoption of AI Features(29:29) Detour and Groupon(37:31) Closing Thoughts Stay in touch: ⁠www.lsvp.com⁠ X: ⁠https://twitter.com/lightspeedvp⁠ LinkedIn: ⁠https://www.linkedin.com/company/lightspeed-venture-partners/⁠ Instagram: ⁠https://www.instagram.com/lightspeedventurepartners/⁠ Subscribe on your favorite podcast app: ⁠generativenow.co⁠ Email: generativenow@lsvp.com The content here does not constitute tax, legal, business or investment advice or an offer to provide such advice, should not be construed as advocating the purchase or sale of any security or investment or a recommendation of any company, and is not an offer, or solicitation of an offer, for the purchase or sale of any security or investment product. For more details please see ⁠lsvp.com/legal⁠.

In this episode, experts discuss a crucial 2024 document outlining appropriate use criteria for multimodality imaging in cardiovascular evaluation before non-emergent non-cardiac surgery, addressing the rising annual surgeries and associated cardiac risks. They delve into balancing the necessity of imaging with cost-effectiveness while exploring the potential of artificial intelligence to enhance future evaluations.

Hey everyone, it's Alex (still traveling!), and oh boy, what a week again! Advanced Voice Mode is finally here from OpenAI, Google updated their Gemini models in a huge way and then Meta announced MultiModal LlaMas and on device mini Llamas (and we also got a "better"? multimodal from Allen AI called MOLMO!)From Weights & Biases perspective, our hackathon was a success this weekend, and then I went down to Menlo Park for my first Meta Connect conference, full of news and updates and will do a full recap here as well. ThursdAI - Recaps of the most high signal AI weekly spaces is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.Overall another crazy week in AI, and it seems that everyone is trying to rush something out the door before OpenAI Dev Day next week (which I'll cover as well!) Get ready, folks, because Dev Day is going to be epic!TL;DR of all topics covered: * Open Source LLMs * Meta llama 3.2 Multimodal models (11B & 90B) (X, HF, try free)* Meta Llama 3.2 tiny models 1B & 3B parameters (X, Blog, download)* Allen AI releases MOLMO - open SOTA multimodal AI models (X, Blog, HF, Try It)* Big CO LLMs + APIs* OpenAI releases Advanced Voice Mode to all & Mira Murati leaves OpenAI * Google updates Gemini 1.5-Pro-002 and 1.5-Flash-002 (Blog)* This weeks Buzz * Our free course is LIVE - more than 3000 already started learning how to build advanced RAG++* Sponsoring tonights AI Tinkerers in Seattle, if you're in Seattle, come through for my demo* Voice & Audio* Meta also launches voice mode (demo)* Tools & Others* Project ORION - holographic glasses are here! (link)Meta gives us new LLaMas and AI hardwareLLama 3.2 Multimodal 11B and 90BThis was by far the biggest OpenSource release of this week (tho see below, may not be the "best"), as a rumored released finally came out, and Meta has given our Llama eyes! Coming with 2 versions (well 4 if you count the base models which they also released), these new MultiModal LLaMas were trained with an adapter architecture, keeping the underlying text models the same, and placing a vision encoder that was trained and finetuned separately on top. LLama 90B is among the best open-source mutlimodal models available— Meta team at launchThese new vision adapters were trained on a massive 6 Billion images, including synthetic data generation by 405B for questions/captions, and finetuned with a subset of 600M high quality image pairs. Unlike the rest of their models, the Meta team did NOT claim SOTA on these models, and the benchmarks are very good but not the best we've seen (Qwen 2 VL from a couple of weeks ago, and MOLMO from today beat it on several benchmarks) With text-only inputs, the Llama 3.2 Vision models are functionally the same as the Llama 3.1 Text models; this allows the Llama 3.2 Vision models to be a drop-in replacement for Llama 3.1 8B/70B with added image understanding capabilities.Seems like these models don't support multi image or video as well (unlike Pixtral for example) nor tool use with images. Meta will also release these models on meta.ai and every other platform, and they cited a crazy 500 million monthly active users of their AI services across all their apps

Join Steve Lehmann & Jeremy Langsam from Portal's Stargaze team on a bimonthly segment of Lab Rats to Unicorns: Rising Stars. In this episode, they explore the groundbreaking work of Zachi Attia, the Director of Artificial Intelligence at Mayo Clinic. With a rich background in electrical engineering and a Ph.D. in Bioinformatics, Zachi discusses his pivotal role in advancing AI models that predict and screen cardiovascular diseases. From his innovative research to real-world applications that are saving lives, this episode offers an inspiring look into the future of healthcare.

Holy s**t, folks! I was off for two weeks, last week OpenAI released GPT-4o-mini and everyone was in my mentions saying, Alex, how are you missing this?? and I'm so glad I missed that last week and not this one, because while GPT-4o-mini is incredible (GPT-4o level distill with incredible speed and almost 99% cost reduction from 2 years ago?) it's not open source. So welcome back to ThursdAI, and buckle up because we're diving into what might just be the craziest week in open-source AI since... well, ever!This week, we saw Meta drop LLAMA 3.1 405B like it's hot (including updated 70B and 8B), Mistral joining the party with their Large V2, and DeepSeek quietly updating their coder V2 to blow our minds. Oh, and did I mention Google DeepMind casually solving math Olympiad problems at silver level medal

In this episode, Lillian Erdahl, MD, FACS is joined by Todd Rosengart, MD, FACS, from the Baylor College of Medicine. They discuss Dr Rosengart's recent article, “Sustaining Lifelong Competency of Surgeons: Multimodality Empowerment Personal and Institutional Strategy,” which focuses on maintaining and ensuring the competency of an aging surgeon workforce. The study provides evidence-based guiding principles as part of a comprehensive “whole of career” strategy that can be adopted at a personal, institutional, and national level.   Disclosure Information: Drs Erdahl and Rosengart have nothing to disclose.   To earn 0.25 AMA PRA Category 1 Credits™ for this episode of the JACS Operative Word Podcast, click here to register for the course and complete the evaluation. Listeners can earn CME credit for this podcast for up to 2 years after the original air date.   Learn more about the Journal of the American College of Surgeons, a monthly peer-reviewed journal publishing original contributions on all aspects of surgery, including scientific articles, collective reviews, experimental investigations, and more.   #JACSOperativeWord

This week on No Priors, Sarah Guo and Elad Gil sit down with Karan Goel and Albert Gu from Cartesia. Karan and Albert first met as Stanford AI Lab PhDs, where their lab invented Space Models or SSMs, a fundamental new primitive for training large-scale foundation models. In 2023, they Founded Cartesia to build real-time intelligence for every device. One year later, Cartesia released Sonic which generates high quality and lifelike speech with a model latency of 135ms—the fastest for a model of this class. Sign up for new podcasts every week. Email feedback to show@no-priors.com Follow us on Twitter: @NoPriorsPod | @Saranormous | @EladGil | @krandiash | @_albertgu Show Notes:  (0:00) Introduction (0:28) Use Cases for Cartesia and Sonic  (1:32) Karan Goel & Albert Gu's professional backgrounds (5:06) Steady State Models (SSMs) versus Transformer Based Architectures  (11:51) Domain Applications for Hybrid Approaches  (13:10) Text to Speech and Voice (17:29) Data, Size of Models and Efficiency  (20:34) Recent Launch of Text to Speech Product (25:01) Multimodality & Building Blocks (25:54) What's Next at Cartesia?  (28:28) Latency in Text to Speech (29:30) Choosing Research Problems Based on Aesthetic  (31:23) Product Demo (32:48) Cartesia Team & Hiring

This conversation explores the different categories of technology available in Australia for clinical treatments with Nadine Dilong, Editor Spa+Clinic. The importance of clinical imaging in the diagnosis and treatment planning process is emphasised. The discussion covers the use of lasers and light therapies for redness and pigmentation, with a focus on the different treatment options and considerations for different skin tones. The conversation also touches on the legislation and safety measures surrounding laser treatments. We discuss skin conditions such as pigmentation, acne and rosacea and offer technology solutions that may accelerate a multi- modality treatment approach.  Key Takeaways Clinical imaging is an important step in the diagnosis and treatment planning process for skin concerns. Laser and light therapies can be effective in treating redness and pigmentation, but the choice of treatment should be based on individual skin type and condition. Legislation and safety measures vary across different states in Australia, highlighting the importance of seeking treatment from qualified practitioners. Regular skincare and clinical imaging can help maintain skin health and prevent future concerns. A correct diagnosis is crucial for determining the appropriate treatment for different types of pigmentation. Multimodality treatments that combine various technologies and devices are becoming more popular in the field of aesthetics. Skincare plays a significant role in preparing the skin for laser and device treatments and maintaining the results. LED light therapy is a simple and effective at-home option for improving skin health. Different lasers and devices can be used to treat acne, acne scars, and pigmentation. CO2, erbium, and fractional lasers are commonly used for skin resurfacing. Laser coring and non-surgical lifting treatments offer alternatives to surgical procedures. Consumers should be cautious when navigating information on social media and seek expert advice. Chapters 00:00 Introduction and the Importance of Clinical Imaging03:16 Diagnosis and Technology in Skin Clinics06:25 Body Composition Analysis07:49 The Importance of Clinical Imaging for Treatment Results09:13 Skincare and Clinical Imaging13:15 Diagnosing Redness and Redness Treatment Options19:11 IPL and Photorejuvenation for Redness23:40 Legislation and Laser Treatment Safety27:38 Laser Treatment for Pigmentation31:45 Different Types of Pigmentation and the Importance of Correct Diagnosis34:24 Multimodality Treatments and the Full Range of Options35:20 The Partnership Between Skincare and Laser/Device Treatments39:08 LED Light Therapy and At-Home Options41:26 Treating Acne with Blue LED and Avicleer Laser45:43 Treating Acne Scars and Pigmentation with Lasers and Devices47:00 Resurfacing the Skin with CO2, Erbium, and Fractional Lasers52:34 Laser Coring and Non-Surgical Lifting Treatments58:25 RF Microneedling and Navigating Information on Social Media Watch the episode here:https://youtu.be/c5_6GcDHvAYSee omnystudio.com/listener for privacy information.

This week, we are joined by Dr. Naima Bhana Lopez, an assistant professor of special education and BCBA-D at Niagara University in New York. Dr. Lopez specializes in enhancing social-communication opportunities for children with developmental disabilities. Her work also explores diversity and equity in special education, empowering natural communication partners to improve outcomes for diverse students and families. With expertise in ABA therapy and special education, Dr. Lopez will be discussing the intersection of these fields with diversity. Download to learn more! Resources  https://www.abainternational.org/media/180194/abai_interprofessional_collaboration_resource_document.pdf   https://aac-learning-center-moodle.psu.edu/   https://abavisualized.com/collections/for-parents/products/aba-en-imagenes-una-guia-visual-para-padres-y-maestros (also available on amazon)   https://abavisualized.com/collections/for-parents/products/aba-visualized-guidebook-2nd-edition (also on Amazon) ................................................................ Autism weekly is now found on all of the major listening apps including apple podcasts, google podcasts, stitcher, Spotify, amazon music, and more. Subscribe to be notified when we post a new podcast. Autism weekly is produced by ABS Kids. ABS Kids is proud to provide diagnostic assessments and ABA therapy to children with developmental delays like Autism Spectrum Disorder. You can learn more about ABS Kids and the Autism Weekly podcast by visiting abskids.com.  

TL;DR: You can now buy tickets, apply to speak, or join the expo for the biggest AI Engineer event of 2024. We're gathering *everyone* you want to meet - see you this June.In last year's the Rise of the AI Engineer we put our money where our mouth was and announced the AI Engineer Summit, which fortunately went well:With ~500 live attendees and over ~500k views online, the first iteration of the AI Engineer industry affair seemed to be well received. Competing in an expensive city with 3 other more established AI conferences in the fall calendar, we broke through in terms of in-person experience and online impact.So at the end of Day 2 we announced our second event: the AI Engineer World's Fair. The new website is now live, together with our new presenting sponsor:We were delighted to invite both Ben Dunphy, co-organizer of the conference and Sam Schillace, the deputy CTO of Microsoft who wrote some of the first Laws of AI Engineering while working with early releases of GPT-4, on the pod to talk about the conference and how Microsoft is all-in on AI Engineering.Rise of the Planet of the AI EngineerSince the first AI Engineer piece, AI Engineering has exploded:and the title has been adopted across OpenAI, Meta, IBM, and many, many other companies:1 year on, it is clear that AI Engineering is not only in full swing, but is an emerging global industry that is successfully bridging the gap:* between research and product, * between general-purpose foundation models and in-context use-cases, * and between the flashy weekend MVP (still great!) and the reliable, rigorously evaluated AI product deployed at massive scale, assisting hundreds of employees and driving millions in profit.The greatly increased scope of the 2024 AI Engineer World's Fair (more stages, more talks, more speakers, more attendees, more expo…) helps us reflect the growth of AI Engineering in three major dimensions:* Global Representation: the 2023 Summit was a mostly-American affair. This year we plan to have speakers from top AI companies across five continents, and explore the vast diversity of approaches to AI across global contexts.* Topic Coverage: * In 2023, the Summit focused on the initial questions that the community wrestled with - LLM frameworks, RAG and Vector Databases, Code Copilots and AI Agents. Those are evergreen problems that just got deeper.* This year the AI Engineering field has also embraced new core disciplines with more explicit focus on Multimodality, Evals and Ops, Open Source Models and GPU/Inference Hardware providers.* Maturity/Production-readiness: Two new tracks are dedicated toward AI in the Enterprise, government, education, finance, and more highly regulated industries or AI deployed at larger scale: * AI in the Fortune 500, covering at-scale production deployments of AI, and* AI Leadership, a closed-door, side event for technical AI leaders to discuss engineering and product leadership challenges as VPs and Heads of AI in their respective orgs.We hope you will join Microsoft and the rest of us as either speaker, exhibitor, or attendee, in San Francisco this June. Contact us with any enquiries that don't fall into the categories mentioned below.Show Notes* Ben Dunphy* 2023 Summit* GitHub confirmed $100m ARR on stage* History of World's Fairs* Sam Schillace* Writely on Acquired.fm* Early Lessons From GPT-4: The Schillace Laws* Semantic Kernel* Sam on Kevin Scott (Microsoft CTO)'s podcast in 2022* AI Engineer World's Fair (SF, Jun 25-27)* Buy Super Early Bird tickets (Listeners can use LATENTSPACE for $100 off any ticket until April 8, or use GROUP if coming in 4 or more)* Submit talks and workshops for Speaker CFPs (by April 8)* Enquire about Expo Sponsorship (Asap.. selling fast)Timestamps* [00:00:16] Intro* [00:01:04] 2023 AI Engineer Summit* [00:03:11] Vendor Neutral* [00:05:33] 2024 AIE World's Fair* [00:07:34] AIE World's Fair: 9 Tracks* [00:08:58] AIE World's Fair Keynotes* [00:09:33] Introducing Sam* [00:12:17] AI in 2020s vs the Cloud in 2000s* [00:13:46] Syntax vs Semantics* [00:14:22] Bill Gates vs GPT-4* [00:16:28] Semantic Kernel and Schillace's Laws of AI Engineering* [00:17:29] Orchestration: Break it into pieces* [00:19:52] Prompt Engineering: Ask Smart to Get Smart* [00:21:57] Think with the model, Plan with Code* [00:23:12] Metacognition vs Stochasticity* [00:24:43] Generating Synthetic Textbooks* [00:26:24] Trade leverage for precision; use interaction to mitigate* [00:27:18] Code is for syntax and process; models are for semantics and intent.* [00:28:46] Hands on AI Leadership* [00:33:18] Multimodality vs "Text is the universal wire protocol"* [00:35:46] Azure OpenAI vs Microsoft Research vs Microsoft AI Division* [00:39:40] On Satya* [00:40:44] Sam at AI Leadership Track* [00:42:05] Final Plug for Tickets & CFPTranscript[00:00:00] Alessio: Hey everyone, welcome to the Latent Space Podcast. This is Alessio, partner and CTO in residence at Decibel Partners, and I'm joined by my co host Swyx, founder of Small[00:00:16] Intro[00:00:16] swyx: AI. Hey, hey, we're back again with a very special episode, this time with two guests and talking about the very in person events rather than online stuff.[00:00:27] swyx: So first I want to welcome Ben Dunphy, who is my co organizer on AI engineer conferences. Hey, hey, how's it going? We have a very special guest. Anyone who's looking at the show notes and the title will preview this later. But I guess we want to set the context. We are effectively doing promo for the upcoming AI Engineer World's Fair that's happening in June.[00:00:49] swyx: But maybe something that we haven't actually recapped much on the pod is just the origin of the AI Engineer Summit and why, what happens and what went down. Ben, I don't know if you'd like to start with the raw numbers that people should have in mind.[00:01:04] 2023 AI Engineer Summit[00:01:04] Ben Dunphy: Yeah, perhaps your listeners would like just a quick background on the summit.[00:01:09] Ben Dunphy: I mean, I'm sure many folks have heard of our events. You know, you launched, we launched the AI Engineer Summit last June with your, your article kind of coining the term that was on the tip of everyone's tongue, but curiously had not been actually coined, which is the term AI Engineer, which is now many people's, Job titles, you know, we're seeing a lot more people come to this event, with the job description of AI engineer, with the job title of AI engineer so, is an event that you and I really talked about since February of 2023, when we met at a hackathon you organized we were both excited by this movement and it hasn't really had a name yet.[00:01:48] Ben Dunphy: We decided that an event was warranted and that's why we move forward with the AI Engineer Summit, which Ended up being a great success. You know, we had over 5, 000 people apply to attend in person. We had over 9, 000 folks attend, online with over 20, 000 on the live stream.[00:02:06] Ben Dunphy: In person, we accepted about 400 attendees and had speakers, workshop instructors and sponsors, all congregating in San Francisco over, two days, um, two and a half days with a, with a welcome reception. So it was quite the event to kick off kind of this movement that's turning into quite an exciting[00:02:24] swyx: industry.[00:02:25] swyx: The overall idea of this is that I kind of view AI engineering, at least in all my work in Latent Space and the other stuff, as starting an industry.[00:02:34] swyx: And I think every industry, every new community, needs a place to congregate. And I definitely think that AI engineer, at least at the conference, is that it's meant to be like the biggest gathering of technical engineering people working with AI. Right. I think we kind of got that spot last year. There was a very competitive conference season, especially in San Francisco.[00:02:54] swyx: But I think as far as I understand, in terms of cultural impact, online impact, and the speakers that people want to see, we, we got them all and it was very important for us to be a vendor neutral type of event. Right. , The reason I partnered with Ben is that Ben has a lot of experience, a lot more experience doing vendor neutral stuff.[00:03:11] Vendor Neutral[00:03:11] swyx: I first met you when I was speaking at one of your events, and now we're sort of business partners on that. And yeah, I mean, I don't know if you have any sort of Thoughts on make, making things vendor neutral, making things more of a community industry conference rather than like something that's owned by one company.[00:03:25] swyx: Yeah.[00:03:25] Ben Dunphy: I mean events that are owned by a company are great, but this is typically where you have product pitches and this smaller internet community. But if you want the truly internet community, if you want a more varied audience and you know, frankly, better content for, especially for a technical audience, you want a vendor neutral event. And this is because when you have folks that are running the event that are focused on one thing and one thing alone, which is quality, quality of content, quality of speakers, quality of the in person experience, and just of general relevance it really elevates everything to the next level.[00:04:01] Ben Dunphy: And when you have someone like yourself who's coming To this content curation the role that you take at this event, and bringing that neutrality with, along with your experience, that really helps to take it to the next level, and then when you have someone like myself, focusing on just the program curation, and the in person experience, then both of our forces combined, we can like, really create this epic event, and so, these vendor neutral events if you've been to a small community event, Typically, these are vendor neutral, but also if you've been to a really, really popular industry event, many of the top industry events are actually vendor neutral.[00:04:37] Ben Dunphy: And that's because of the fact that they're vendor neutral, not in spite of[00:04:41] swyx: it. Yeah, I've been pretty open about the fact that my dream is to build the KubeCon of AI. So if anyone has been in the Kubernetes world, they'll understand what that means. And then, or, or instead of the NeurIPS, NeurIPS for engineers, where engineers are the stars and engineers are sharing their knowledge.[00:04:57] swyx: Perspectives, because I think AI is definitely moving over from research to engineering and production. I think one of my favorite parts was just honestly having GitHub and Microsoft support, which we'll cover in a bit, but you know, announcing finally that GitHub's copilot was such a commercial success I think was the first time that was actually confirmed by anyone in public.[00:05:17] swyx: For me, it's also interesting as sort of the conference curator to put Microsoft next to competitors some of which might be much smaller AI startups and to see what, where different companies are innovating in different areas.[00:05:27] swyx: Well, they're next to[00:05:27] Ben Dunphy: each other in the arena. So they can be next to each other on stage too.[00:05:33] Why AIE World's Fair[00:05:33] swyx: Okay, so this year World's Fair we are going a lot bigger what details are we disclosing right now? Yeah,[00:05:39] Ben Dunphy: I guess we should start with the name why are we calling it the World's Fair? And I think we need to go back to what inspired this, what actually the original World's Fair was, which was it started in the late 1700s and went to the early 1900s.[00:05:53] Ben Dunphy: And it was intended to showcase the incredible achievements. Of nation states, corporations, individuals in these grand expos. So you have these miniature cities actually being built for these grand expos. In San Francisco, for example, you had the entire Marina District built up in absolutely new construction to showcase the achievements of industry, architecture, art, and culture.[00:06:16] Ben Dunphy: And many of your listeners will know that in 1893, the Nikola Tesla famously provided power to the Chicago World's Fair with his 8 seat power generator. There's lots of great movies and documentaries about this. That was the first electric World's Fair, which thereafter it was referred to as the White City.[00:06:33] Ben Dunphy: So in today's world we have technological change that's similar to what was experienced during the industrial revolution in how it's, how it's just upending our entire life, how we live, work, and play. And so we have artificial intelligence, which has long been the dream of humanity.[00:06:51] Ben Dunphy: It's, it's finally here. And the pace of technological change is just accelerating. So with this event, as you mentioned, we, we're aiming to create a singular event where the world's foremost experts, builders, and practitioners can come together to exchange and reflect. And we think this is not only good for business, but it's also good for our mental health.[00:07:12] Ben Dunphy: It slows things down a bit from the Twitter news cycle to an in person festival of smiles, handshakes, connections, and in depth conversations that online media and online events can only ever dream of replicating. So this is an expo led event where the world's top companies will mingle with the world's top founders and AI engineers who are building and enhanced by AI.[00:07:34] AIE World's Fair: 9 Tracks[00:07:34] Ben Dunphy: And not to mention, we're featuring over a hundred talks and workshops across[00:07:37] swyx: nine tracks. Yeah, I mean, those nine tracks will be fun. Actually, do we have a little preview of the tracks in the, the speakers?[00:07:43] Ben Dunphy: We do. Folks can actually see them today at our website. We've updated that at ai.[00:07:48] Ben Dunphy: engineer. So we'd encourage them to go there to see that. But for those just listening, we have nine tracks. So we have multimodality. We have retrieval augmented generation. Featuring LLM frameworks and vector databases, evals and LLM ops, open source models, code gen and dev tools, GPUs and inference, AI agent applications, AI in the fortune 500, and then we have a special track for AI leadership which you can access by purchasing the VP pass which is different from the, the other passes we have.[00:08:20] Ben Dunphy: And I won't go into the Each of these tracks in depth, unless you want to, Swyx but there's more details on the website at ai. engineer.[00:08:28] swyx: I mean, I, I, very much looking forward to talking to our special guests for the last track, I think, which is the what a lot of yeah, leaders are thinking about, which is how to, Inspire innovation in their companies, especially the sort of larger organizations that might not have the in house talents for that kind of stuff.[00:08:47] swyx: So yeah, we can talk about the expo, but I'm very keen to talk about the presenting sponsor if you want to go slightly out of order from our original plan.[00:08:58] AIE World's Fair Keynotes[00:08:58] Ben Dunphy: Yeah, absolutely. So you know, for the stage of keynotes, we have talks confirmed from Microsoft, OpenAI, AWS, and Google.[00:09:06] Ben Dunphy: And our presenting sponsor is joining the stage with those folks. And so that presenting sponsor this year is a dream sponsor. It's Microsoft. It's the company really helping to lead the charge. And into this wonderful new era that we're all taking part in. So, yeah,[00:09:20] swyx: you know, a bit of context, like when we first started planning this thing, I was kind of brainstorming, like, who would we like to get as the ideal presenting sponsors, as ideal partners long term, just in terms of encouraging the AI engineering industry, and it was Microsoft.[00:09:33] Introducing Sam[00:09:33] swyx: So Sam, I'm very excited to welcome you onto the podcast. You are CVP and Deputy CTO of Microsoft. Welcome.[00:09:40] Sam Schillace: Nice to be here. I'm looking forward to, I was looking for, to Lessio saying my last name correctly this time. Oh[00:09:45] swyx: yeah. So I, I studiously avoided saying, saying your last name, but apparently it's an Italian last name.[00:09:50] swyx: Ski Lache. Ski[00:09:51] Alessio: Lache. Yeah. No, that, that's great, Sean. That's great as a musical person.[00:09:54] swyx: And it, it's also, yeah, I pay attention to like the, the, the lilt. So it's ski lache and the, the slow slowing of the law is, is what I focused[00:10:03] Sam Schillace: on. You say both Ls. There's no silent letters, you say[00:10:07] Alessio: both of those. And it's great to have you, Sam.[00:10:09] Alessio: You know, we've known each other now for a year and a half, two years, and our first conversation, well, it was at Lobby Conference, and then we had a really good one in the kind of parking lot of a Safeway, because we didn't want to go into Starbucks to meet, so we sat outside for about an hour, an hour and a half, and then you had to go to a Bluegrass concert, so it was great.[00:10:28] Alessio: Great meeting, and now, finally, we have you on Lanespace.[00:10:31] Sam Schillace: Cool, cool. Yeah, I'm happy to be here. It's funny, I was just saying to Swyx before you joined that, like, it's kind of an intimidating podcast. Like, when I listen to this podcast, it seems to be, like, one of the more intelligent ones, like, more, more, like, deep technical folks on it.[00:10:44] Sam Schillace: So, it's, like, it's kind of nice to be here. It's fun. Bring your A game. Hopefully I'll, I'll bring mine. I[00:10:49] swyx: mean, you've been programming for longer than some of our listeners have been alive, so I don't think your technical chops are in any doubt. So you were responsible for Rightly as one of your early wins in your career, which then became Google Docs, and obviously you were then responsible for a lot more G Suite.[00:11:07] swyx: But did you know that you covered in Acquired. fm episode 9, which is one of the podcasts that we model after.[00:11:13] Sam Schillace: Oh, cool. I didn't, I didn't realize that the most fun way to say this is that I still have to this day in my personal GDocs account, the very first Google doc, like I actually have it.[00:11:24] Sam Schillace: And I looked it up, like it occurred to me like six months ago that it was probably around and I went and looked and it's still there. So it's like, and it's kind of a funny thing. Cause it's like the backend has been rewritten at least twice that I know of the front end has been re rewritten at least twice that I know of.[00:11:38] Sam Schillace: So. I'm not sure what sense it's still the original one it's sort of more the idea of the original one, like the NFT of it would probably be more authentic. I[00:11:46] swyx: still have it. It's a ship athesia thing. Does it, does it say hello world or something more mundane?[00:11:52] Sam Schillace: It's, it's, it's me and Steve Newman trying to figure out if some collaboration stuff is working, and also a picture of Edna from the Incredibles that I probably pasted in later, because that's That's too early for that, I think.[00:12:05] swyx: People can look up your LinkedIn, and we're going to link it on the show notes, but you're also SVP of engineering for Box, and then you went back to Google to do Google, to lead Google Maps, and now you're deputy CTO.[00:12:17] AI in 2020s vs the Cloud in 2000s[00:12:17] swyx: I mean, there's so many places to start, but maybe one place I like to start off with is do you have a personal GPT 4 experience.[00:12:25] swyx: Obviously being at Microsoft, you have, you had early access and everyone talks about Bill Gates's[00:12:30] Sam Schillace: demo. Yeah, it's kind of, yeah, that's, it's kind of interesting. Like, yeah, we got access, I got access to it like in September of 2022, I guess, like before it was really released. And I it like almost instantly was just like mind blowing to me how good it was.[00:12:47] Sam Schillace: I would try experiments like very early on, like I play music. There's this thing called ABC notation. That's like an ASCII way to represent music. And like, I was like, I wonder if it can like compose a fiddle tune. And like it composed a fiddle tune. I'm like, I wonder if it can change key, change the key.[00:13:01] Sam Schillace: Like it's like really, it was like very astonishing. And I sort of, I'm very like abstract. My background is actually more math than CS. I'm a very abstract thinker and sort of categorical thinker. And the, the thing that occurred to me with, with GPT 4 the first time I saw it was. This is really like the beginning, it's the beginning of V2 of the computer industry completely.[00:13:23] Sam Schillace: I had the same feeling I had when, of like a category shifting that I had when the cloud stuff happened with the GDocs stuff, right? Where it's just like, all of a sudden this like huge vista opens up of capabilities. And I think the way I characterized it, which is a little bit nerdy, but I'm a nerd so lean into it is like everything until now has been about syntax.[00:13:46] Syntax vs Semantics[00:13:46] Sam Schillace: Like, we have to do mediation. We have to describe the real world in forms that the digital world can manage. And so we're the mediation, and we, like, do that via things like syntax and schema and programming languages. And all of a sudden, like, this opens the door to semantics, where, like, you can express intention and meaning and nuance and fuzziness.[00:14:04] Sam Schillace: And the machine itself is doing, the model itself is doing a bunch of the mediation for you. And like, that's obviously like complicated. We can talk about the limits and stuff, and it's getting better in some ways. And we're learning things and all kinds of stuff is going on around it, obviously.[00:14:18] Sam Schillace: But like, that was my immediate reaction to it was just like, Oh my God.[00:14:22] Bill Gates vs GPT-4[00:14:22] Sam Schillace: Like, and then I heard about the build demo where like Bill had been telling Kevin Scott this, This investment is a waste. It's never going to work. AI is blah, blah, blah. And come back when it can pass like an AP bio exam.[00:14:33] Sam Schillace: And they actually literally did that at one point, they brought in like the world champion of the, like the AP bio test or whatever the AP competition and like it and chat GPT or GPT 4 both did the AP bio and GPT 4 beat her. So that was the moment that convinced Bill that this was actually real.[00:14:53] Sam Schillace: Yeah, it's fun. I had a moment with him actually about three weeks after that when we had been, so I started like diving in on developer tools almost immediately and I built this thing with a small team that's called the Semantic Kernel which is one of the very early orchestrators just because I wanted to be able to put code and And inference together.[00:15:10] Sam Schillace: And that's probably something we should dig into more deeply. Cause I think there's some good insights in there, but I I had a bunch of stuff that we were building and then I was asked to go meet with Bill Gates about it and he's kind of famously skeptical and, and so I was a little bit nervous to meet him the first time.[00:15:25] Sam Schillace: And I started the conversation with, Hey, Bill, like three weeks ago, you would have called BS on everything I'm about to show you. And I would probably have agreed with you, but we've both seen this thing. And so we both know it's real. So let's skip that part and like, talk about what's possible.[00:15:39] Sam Schillace: And then we just had this kind of fun, open ended conversation and I showed him a bunch of stuff. So that was like a really nice, fun, fun moment as well. Well,[00:15:46] swyx: that's a nice way to meet Bill Gates and impress[00:15:48] Sam Schillace: him. A little funny. I mean, it's like, I wasn't sure what he would think of me, given what I've done and his.[00:15:54] Sam Schillace: Crown Jewel. But he was nice. I think he likes[00:15:59] swyx: GDocs. Crown Jewel as in Google Docs versus Microsoft Word? Office.[00:16:03] Sam Schillace: Yeah. Yeah, versus Office. Yeah, like, I think, I mean, I can imagine him not liking, I met Steven Snofsky once and he sort of respectfully, but sort of grimaced at me. You know, like, because of how much trauma I had caused him.[00:16:18] Sam Schillace: So Bill was very nice to[00:16:20] swyx: me. In general it's like friendly competition, right? They keep you, they keep you sharp, you keep each[00:16:24] Sam Schillace: other sharp. Yeah, no, I think that's, it's definitely respect, it's just kind of funny.[00:16:28] Semantic Kernel and Schillace's Laws of AI Engineering[00:16:28] Sam Schillace: Yeah,[00:16:28] swyx: So, speaking of semantic kernel, I had no idea that you were that deeply involved, that you actually had laws named after you.[00:16:35] swyx: This only came up after looking into you for a little bit. Skelatches laws, how did those, what's the, what's the origin[00:16:41] Sam Schillace: story? Hey! Yeah, that's kind of funny. I'm actually kind of a modest person and so I'm sure I feel about having my name attached to them. Although I do agree with all, I believe all of them because I wrote all of them.[00:16:49] Sam Schillace: This is like a designer, John Might, who works with me, decided to stick my name on them and put them out there. Seriously, but like, well, but like, so this was just I, I'm not, I don't build models. Like I'm not an AI engineer in the sense of, of like AI researcher that's like doing inference. Like I'm somebody who's like consuming the models.[00:17:09] Sam Schillace: Exactly. So it's kind of funny when you're talking about AI engineering, like it's a good way of putting it. Cause that's how like I think about myself. I'm like, I'm an app builder. I just want to build with this tool. Yep. And so we spent all of the fall and into the winter in that first year, like Just trying to build stuff and learn how this tool worked.[00:17:29] Orchestration: Break it into pieces[00:17:29] Sam Schillace: And I guess those are a little bit in the spirit of like Robert Bentley's programming pearls or something. I was just like, let's kind of distill some of these ideas down of like. How does this thing work? I saw something I still see today with people doing like inference is still kind of expensive.[00:17:46] Sam Schillace: GPUs are still kind of scarce. And so people try to get everything done in like one shot. And so there's all this like prompt tuning to get things working. And one of the first laws was like, break it into pieces. Like if it's hard for you, it's going to be hard for the model. But if it's you know, there's this kind of weird thing where like, it's.[00:18:02] Sam Schillace: It's absolutely not a human being, but starting to think about, like, how would I solve the problem is often a good way to figure out how to architect the program so that the model can solve the problem. So, like, that was one of the first laws. That came from me just trying to, like, replicate a test of a, like, a more complicated, There's like a reasoning process that you have to go through that, that Google was, was the react, the react thing, and I was trying to get GPT 4 to do it on its own.[00:18:32] Sam Schillace: And, and so I'd ask it the question that was in this paper, and the answer to the question is like the year 2000. It's like, what year did this particular author who wrote this book live in this country? And you've kind of got to carefully reason through it. And like, I could not get GPT 4 to Just to answer the question with the year 2000.[00:18:50] Sam Schillace: And if you're thinking about this as like the kernel is like a pipelined orchestrator, right? It's like very Unix y, where like you have a, some kind of command and you pipe stuff to the next parameters and output to the next thing. So I'm thinking about this as like one module in like a pipeline, and I just want it to give me the answer.[00:19:05] Sam Schillace: I don't want anything else. And I could not prompt engineer my way out of that. I just like, it was giving me a paragraph or reasoning. And so I sort of like anthropomorphized a little bit and I was like, well, the only way you can think about stuff is it can think out loud because there's nothing else that the model does.[00:19:19] Sam Schillace: It's just doing token generation. And so it's not going to be able to do this reasoning if it can't think out loud. And that's why it's always producing this. But if you take that paragraph of output, which did get to the right answer and you pipe it into a second prompt. That just says read this conversation and just extract the answer and report it back.[00:19:38] Sam Schillace: That's an easier task. That would be an easier task for you to do or me to do. It's easier reasoning. And so it's an easier thing for the model to do and it's much more accurate. And that's like 100 percent accurate. It always does that. So like that was one of those, those insights on the that led to the, the choice loss.[00:19:52] Prompt Engineering: Ask Smart to Get Smart[00:19:52] Sam Schillace: I think one of the other ones that's kind of interesting that I think people still don't fully appreciate is that GPT 4 is the rough equivalent of like a human being sitting down for centuries or millennia and reading all the books that they can find. It's this vast mind, right, and the embedding space, the latent space, is 100, 000 K, 100, 000 dimensional space, right?[00:20:14] Sam Schillace: Like it's this huge, high dimensional space, and we don't have good, um, Intuition about high dimensional spaces, like the topology works in really weird ways, connectivity works in weird ways. So a lot of what we're doing is like aiming the attention of a model into some part of this very weirdly connected space.[00:20:30] Sam Schillace: That's kind of what prompt engineering is. But that kind of, like, what we observed to begin with that led to one of those laws was You know, ask smart to get smart. And I think we've all, we all understand this now, right? Like this is the whole field of prompt engineering. But like, if you ask like a simple, a simplistic question of the model, you'll get kind of a simplistic answer.[00:20:50] Sam Schillace: Cause you're pointing it at a simplistic part of that high dimensional space. And if you ask it a more intelligent question, you get more intelligent stuff back out. And so I think that's part of like how you think about programming as well. It's like, how are you directing the attention of the model?[00:21:04] Sam Schillace: And I think we still don't have a good intuitive feel for that. To me,[00:21:08] Alessio: the most interesting thing is how do you tie the ask smart, get smart with the syntax and semantics piece. I gave a talk at GDC last week about the rise of full stack employees and how these models are like semantic representation of tasks that people do.[00:21:23] Alessio: But at the same time, we have code. Also become semantic representation of code. You know, I give you the example of like Python that sort it's like really a semantic function. It's not code, but it's actually code underneath. How do you think about tying the two together where you have code?[00:21:39] Alessio: To then extract the smart parts so that you don't have to like ask smart every time and like kind of wrap them in like higher level functions.[00:21:46] Sam Schillace: Yeah, this is, this is actually, we're skipping ahead to kind of later in the conversation, but I like to, I usually like to still stuff down in these little aphorisms that kind of help me remember them.[00:21:57] Think with the model, Plan with Code[00:21:57] Sam Schillace: You know, so we can dig into a bunch of them. One of them is pixels are free, one of them is bots are docs. But the one that's interesting here is Think with the model, plan with code. And so one of the things, so one of the things we've realized, we've been trying to do lots of these like longer running tasks.[00:22:13] Sam Schillace: Like we did this thing called the infinite chatbot, which was the successor to the semantic kernel, which is an internal project. It's a lot like GPTs. The open AI GPT is, but it's like a little bit more advanced in some ways, kind of deep exploration of a rag based bot system. And then we did multi agents from that, trying to do some autonomy stuff and we're, and we're kind of banging our head against this thing.[00:22:34] Sam Schillace: And you know, one of the things I started to realize, this is going to get nerdy for a second. I apologize, but let me dig in on it for just a second. No apology needed. Um, we realized is like, again, this is a little bit of an anthropomorphism and an illusion that we're having. So like when we look at these models, we think there's something continuous there.[00:22:51] Sam Schillace: We're having a conversation with chat GPT or whatever with Azure open air or like, like what's really happened. It's a little bit like watching claymation, right? Like when you watch claymation, you don't think that the model is actually the clay model is actually really alive. You know, that there's like a bunch of still disconnected slot screens that your mind is connecting into a continuous experience.[00:23:12] Metacognition vs Stochasticity[00:23:12] Sam Schillace: And that's kind of the same thing that's going on with these models. Like they're all the prompts are disconnected no matter what. Which means you're putting a lot of weight on memory, right? This is the thing we talked about. You're like, you're putting a lot of weight on precision and recall of your memory system.[00:23:27] Sam Schillace: And so like, and it turns out like, because the models are stochastic, they're kind of random. They'll make stuff up if things are missing. If you're naive about your, your memory system, you'll get lots of like accumulated similar memories that will kind of clog the system, things like that. So there's lots of ways in which like, Memory is hard to manage well, and, and, and that's okay.[00:23:47] Sam Schillace: But what happens is when you're doing plans and you're doing these longer running things that you're talking about, that second level, the metacognition is very vulnerable to that stochastic noise, which is like, I totally want to put this on a bumper sticker that like metacognition is susceptible to stochasticity would be like the great bumper sticker.[00:24:07] Sam Schillace: So what, these things are very vulnerable to feedback loops when they're trying to do autonomy, and they're very vulnerable to getting lost. So we've had these, like, multi agent Autonomous agent things get kind of stuck on like complimenting each other, or they'll get stuck on being quote unquote frustrated and they'll go on strike.[00:24:22] Sam Schillace: Like there's all kinds of weird like feedback loops you get into. So what we've learned to answer your question of how you put all this stuff together is You have to, the model's good at thinking, but it's not good at planning. So you do planning in code. So you have to describe the larger process of what you're doing in code somehow.[00:24:38] Sam Schillace: So semantic intent or whatever. And then you let the model kind of fill in the pieces.[00:24:43] Generating Synthetic Textbooks[00:24:43] Sam Schillace: I'll give a less abstract like example. It's a little bit of an old example. I did this like last year, but at one point I wanted to see if I could generate textbooks. And so I wrote this thing called the textbook factory.[00:24:53] Sam Schillace: And it's, it's tiny. It's like a Jupyter notebook with like. You know, 200 lines of Python and like six very short prompts, but what you basically give it a sentence. And it like pulls out the topic and the level of, of, from that sentence, so you, like, I would like fifth grade reading. I would like eighth grade English.[00:25:11] Sam Schillace: His English ninth grade, US history, whatever. That by the way, all, all by itself, like would've been an almost impossible job like three years ago. Isn't, it's like totally amazing like that by itself. Just parsing an arbitrary natural language sentence to get these two pieces of information out is like almost trivial now.[00:25:27] Sam Schillace: Which is amazing. So it takes that and it just like makes like a thousand calls to the API and it goes and builds a full year textbook, like decides what the curriculum is with one of the prompts. It breaks it into chapters. It writes all the lessons and lesson plans and like builds a teacher's guide with all the answers to all the questions.[00:25:42] Sam Schillace: It builds a table of contents, like all that stuff. It's super reliable. You always get a textbook. It's super brittle. You never get a cookbook or a novel like but like you could kind of define that domain pretty care, like I can describe. The metacognition, the high level plan for how do you write a textbook, right?[00:25:59] Sam Schillace: You like decide the curriculum and then you write all the chapters and you write the teacher's guide and you write the table content, like you can, you can describe that out pretty well. And so having that like code exoskeleton wrapped around the model is really helpful, like it keeps the model from drifting off and then you don't have as many of these vulnerabilities around memory that you would normally have.[00:26:19] Sam Schillace: So like, that's kind of, I think where the syntax and semantics comes together right now.[00:26:24] Trade leverage for precision; use interaction to mitigate[00:26:24] Sam Schillace: And then I think the question for all of us is. How do you get more leverage out of that? Right? So one of the things that I don't love about virtually everything anyone's built for the last year and a half is people are holding the hands of the model on everything.[00:26:37] Sam Schillace: Like the leverage is very low, right? You can't turn. These things loose to do anything really interesting for very long. You can kind of, and the places where people are getting more work out per unit of work in are usually where somebody has done exactly what I just described. They've kind of figured out what the pattern of the problem is in enough of a way that they can write some code for it.[00:26:59] Sam Schillace: And then that that like, so I've seen like sales support stuff. I've seen like code base tuning stuff of like, there's lots of things that people are doing where like, you can get a lot of value in some relatively well defined domain using a little bit of the model's ability to think for you and a little, and a little bit of code.[00:27:18] Code is for syntax and process; models are for semantics and intent.[00:27:18] Sam Schillace: And then I think the next wave is like, okay, do we do stuff like domain specific languages to like make the planning capabilities better? Do we like start to build? More sophisticated primitives. We're starting to think about and talk about like power automate and a bunch of stuff inside of Microsoft that we're going to wrap in these like building blocks.[00:27:34] Sam Schillace: So the models have these chunks of reliable functionality that they can invoke as part of these plans, right? Because you don't want like, if you're going to ask the model to go do something and the output's going to be a hundred thousand lines of code, if it's got to generate that code every time, the randomness, the stochasticity is like going to make that basically not reliable.[00:27:54] Sam Schillace: You want it to generate it like a 10 or 20 line high level semantic plan for this thing that gets handed to some markup executor that runs it and that invokes that API, that 100, 000 lines of code behind it, API call. And like, that's a really nice robust system for now. And then as the models get smarter as new models emerge, then we get better plans, we get more sophistication.[00:28:17] Sam Schillace: In terms of what they can choose, things like that. Right. So I think like that feels like that's probably the path forward for a little while, at least, like there was, there was a lot there. I, sorry, like I've been thinking, you can tell I've been thinking about it a lot. Like this is kind of all I think about is like, how do you build.[00:28:31] Sam Schillace: Really high value stuff out of this. And where do we go? Yeah. The, the role where[00:28:35] swyx: we are. Yeah. The intermixing of code and, and LMS is, is a lot of the role of the AI engineer. And I, I, I think in a very real way, you were one of the first to, because obviously you had early access. Honestly, I'm surprised.[00:28:46] Hands on AI Leadership[00:28:46] swyx: How are you so hands on? How do you choose to, to dedicate your time? How do you advise other tech leaders? Right. You know, you, you are. You have people working for you, you could not be hands on, but you seem to be hands on. What's the allocation that people should have, especially if they're senior tech[00:29:03] Sam Schillace: leaders?[00:29:04] Sam Schillace: It's mostly just fun. Like, I'm a maker, and I like to build stuff. I'm a little bit idiosyncratic. I I've got ADHD, and so I won't build anything. I won't work on anything I'm bored with. So I have no discipline. If I'm not actually interested in the thing, I can't just, like, do it, force myself to do it.[00:29:17] Sam Schillace: But, I mean, if you're not interested in what's going on right now in the industry, like, go find a different industry, honestly. Like, I seriously, like, this is, I, well, it's funny, like, I don't mean to be snarky, but, like, I was at a dinner, like, a, I don't know, six months ago or something, And I was sitting next to a CTO of a large, I won't name the corporation because it would name the person, but I was sitting next to the CTO of a very large Japanese technical company, and he was like, like, nothing has been interesting since the internet, and this is interesting now, like, this is fun again.[00:29:46] Sam Schillace: And I'm like, yeah, totally, like this is like, the most interesting thing that's happened in 35 years of my career, like, we can play with semantics and natural language, and we can have these things that are like sort of active, can kind of be independent in certain ways and can do stuff for us and can like, reach all of these interesting problems.[00:30:02] Sam Schillace: So like that's part of it of it's just kind of fun to, to do stuff and to build stuff. I, I just can't, can't resist. I'm not crazy hands-on, like, I have an eng like my engineering team's listening right now. They're like probably laughing 'cause they, I never, I, I don't really touch code directly 'cause I'm so obsessive.[00:30:17] Sam Schillace: I told them like, if I start writing code, that's all I'm gonna do. And it's probably better if I stay a little bit high level and like, think about. I've got a really great couple of engineers, a bunch of engineers underneath me, a bunch of designers underneath me that are really good folks that we just bounce ideas off of back and forth and it's just really fun.[00:30:35] Sam Schillace: That's the role I came to Microsoft to do, really, was to just kind of bring some energy around innovation, some energy around consumer, We didn't know that this was coming when I joined. I joined like eight months before it hit us, but I think Kevin might've had an idea it was coming. And and then when it hit, I just kind of dove in with both feet cause it's just so much fun to do.[00:30:55] Sam Schillace: Just to tie it back a little bit to the, the Google Docs stuff. When we did rightly originally the world it's not like I built rightly in jQuery or anything. Like I built that thing on bare metal back before there were decent JavaScript VMs.[00:31:10] Sam Schillace: I was just telling somebody today, like you were rate limited. So like just computing the diff when you type something like doing the string diff, I had to write like a binary search on each end of the string diff because like you didn't have enough iterations of a for loop to search character by character.[00:31:24] Sam Schillace: I mean, like that's how rough it was none of the browsers implemented stuff directly, whatever. It's like, just really messy. And like, that's. Like, as somebody who's been doing this for a long time, like, that's the place where you want to engage, right? If things are easy, and it's easy to go do something, it's too late.[00:31:42] Sam Schillace: Even if it's not too late, it's going to be crowded, but like the right time to do something new and disruptive and technical is, first of all, still when it's controversial, but second of all, when you have this, like, you can see the future, you ask this, like, what if question, and you can see where it's going, But you have this, like, pit in your stomach as an engineer as to, like, how crappy this is going to be to do.[00:32:04] Sam Schillace: Like, that's really the right moment to engage with stuff. We're just like, this is going to suck, it's going to be messy, I don't know what the path is, I'm going to get sticks and thorns in my hair, like I, I, it's going to have false starts, and I don't really, I'm going to This is why those skeletchae laws are kind of funny, because, like, I, I, like You know, I wrote them down at one point because they were like my best guess, but I'm like half of these are probably wrong, and I think they've all held up pretty well, but I'm just like guessing along with everybody else, we're just trying to figure this thing out still, right, and like, and I think the only way to do that is to just engage with it.[00:32:34] Sam Schillace: You just have to like, build stuff. If you're, I can't tell you the number of execs I've talked to who have opinions about AI and have not sat down with anything for more than 10 minutes to like actually try to get anything done. You know, it's just like, it's incomprehensible to me that you can watch this stuff through the lens of like the press and forgive me, podcasts and feel like you actually know what you're talking about.[00:32:59] Sam Schillace: Like, you have to like build stuff. Like, break your nose on stuff and like figure out what doesn't work.[00:33:04] swyx: Yeah, I mean, I view us as a starting point, as a way for people to get exposure on what we're doing. They should be looking at, and they still have to do the work as do we. Yeah, I'll basically endorse, like, I think most of the laws.[00:33:18] Multimodality vs "Text is the universal wire protocol"[00:33:18] swyx: I think the one I question the most now is text is the universal wire protocol. There was a very popular article, a text that used a universal interface by Rune who now works at OpenAI. And I, actually, we just, we just dropped a podcast with David Luan, who's CEO of Adept now, but he was VP of Eng, and he pitched Kevin Scott for the original Microsoft investment in OpenAI.[00:33:40] swyx: Where he's basically pivoting to or just betting very hard on multimodality. I think that's something that we don't really position very well. I think this year, we're trying to all figure it out. I don't know if you have an updated perspective on multi modal models how that affects agents[00:33:54] Sam Schillace: or not.[00:33:55] Sam Schillace: Yeah, I mean, I think the multi I think multi modality is really important. And I, I think it's only going to get better from here. For sure. Yeah, the text is the universal wire protocol. You're probably right. Like, I don't know that I would defend that one entirely. Note that it doesn't say English, right?[00:34:09] Sam Schillace: Like it's, it's not, that's even natural language. Like there's stuff like Steve Luko, who's the guy who created TypeScript, created TypeChat, right? Which is this like way to get LLMs to be very precise and return syntax and correct JavaScript. So like, I, yeah, I think like multimodality, like, I think part of the challenge with it is like, it's a little harder to access.[00:34:30] Sam Schillace: Programatically still like I think you know and I do think like, You know like when when like dahly and stuff started to come Out I was like, oh photoshop's in trouble cuz like, you know I'm just gonna like describe images And you don't need photos of Photoshop anymore Which hasn't played out that way like they're actually like adding a bunch of tools who look like you want to be able to you know for multimodality be really like super super charged you need to be able to do stuff like Descriptively, like, okay, find the dog in this picture and mask around it.[00:34:58] Sam Schillace: Okay, now make it larger and whatever. You need to be able to interact with stuff textually, which we're starting to be able to do. Like, you can do some of that stuff. But there's probably a whole bunch of new capabilities that are going to come out that are going to make it more interesting.[00:35:11] Sam Schillace: So, I don't know, like, I suspect we're going to wind up looking kind of like Unix at the end of the day, where, like, there's pipes and, like, Stuff goes over pipes, and some of the pipes are byte character pipes, and some of them are byte digital or whatever like binary pipes, and that's going to be compatible with a lot of the systems we have out there, so like, that's probably still And I think there's a lot to be gotten from, from text as a language, but I suspect you're right.[00:35:37] Sam Schillace: Like that particular law is not going to hold up super well. But we didn't have multimodal going when I wrote it. I'll take one out as well.[00:35:46] Azure OpenAI vs Microsoft Research vs Microsoft AI Division[00:35:46] swyx: I know. Yeah, I mean, the innovations that keep coming out of Microsoft. You mentioned multi agent. I think you're talking about autogen.[00:35:52] swyx: But there's always research coming out of MSR. Yeah. PHY1, PHY2. Yeah, there's a bunch of[00:35:57] Sam Schillace: stuff. Yeah.[00:35:59] swyx: What should, how should the outsider or the AI engineer just as a sort of final word, like, How should they view the Microsoft portfolio things? I know you're not here to be a salesman, but What, how do you explain You know, Microsoft's AI[00:36:12] Sam Schillace: work to people.[00:36:13] Sam Schillace: There's a lot of stuff going on. Like, first of all, like, I should, I'll be a little tiny bit of a salesman for, like, two seconds and just point out that, like, one of the things we have is the Microsoft for Startups Founders Hub. So, like, you can get, like, Azure credits and stuff from us. Like, up to, like, 150 grand, I think, over four years.[00:36:29] Sam Schillace: So, like, it's actually pretty easy to get. Credit you can start, I 500 bucks to start or something with very little other than just an idea. So like there's, that's pretty cool. Like, I like Microsoft is very much all in on AI at, at many levels. And so like that, you mentioned, you mentioned Autogen, like, So I sit in the office of the CTO, Microsoft Research sits under him, under the office of the CTO as well.[00:36:51] Sam Schillace: So the Autogen group came out of somebody in MSR, like in that group. So like there's sort of. The spectrum of very researchy things going on in research, where we're doing things like Phi, which is the small language model efficiency exploration that's really, really interesting. Lots of very technical folks there that are building different kinds of models.[00:37:10] Sam Schillace: And then there's like, groups like my group that are kind of a little bit in the middle that straddle product and, and, and research and kind of have a foot in both worlds and are trying to kind of be a bridge into the product world. And then there's like a whole bunch of stuff on the product side of things.[00:37:23] Sam Schillace: So there's. All the Azure OpenAI stuff, and then there's all the stuff that's in Office and Windows. And I, so I think, like, the way, I don't know, the way to think about Microsoft is we're just powering AI at every level we can, and making it as accessible as we can to both end users and developers.[00:37:42] Sam Schillace: There's this really nice research arm at one end of that spectrum that's really driving the cutting edge. The fee stuff is really amazing. It broke the chinchella curves. Right, like we didn't, that's the textbooks are all you need paper, and it's still kind of controversial, but like that was really a surprising result that came out of MSR.[00:37:58] Sam Schillace: And so like I think Microsoft is both being a thought leader on one end, on the other end with all the Azure OpenAI, all the Azure tooling that we have, like very much a developer centric, kind of the tinkerer's paradise that Microsoft always was. It's like a great place to come and consume all these things.[00:38:14] Sam Schillace: There's really amazing stuff ideas that we've had, like these very rich, long running, rag based chatbots that we didn't talk about that are like now possible to just go build with Azure AI Studio for yourself. You can build and deploy like a chatbot that's trained on your data specifically, like very easily and things like that.[00:38:31] Sam Schillace: So like there's that end of things. And then there's all this stuff that's in Office, where like, you could just like use the copilots both in Bing, but also just like daily your daily work. So like, it's just kind of everywhere at this point, like everyone in the company thinks about it all the time.[00:38:43] Sam Schillace: There's like no single answer to that question. That was way more salesy than I thought I was capable of, but like, that is actually the genuine truth. Like, it is all the time, it is all levels, it is all the way from really pragmatic, approachable stuff for somebody starting out who doesn't know things, all the way to like Absolutely cutting edge research, silicon, models, AI for science, like, we didn't talk about any of the AI for science stuff, I've seen magical stuff coming out of the research group on that topic, like just crazy cool stuff that's coming, so.[00:39:13] Sam Schillace: You've[00:39:14] swyx: called this since you joined Microsoft. I point listeners to the podcast that you did in 2022, pre ChatGBT with Kevin Scott. And yeah, you've been saying this from the beginning. So this is not a new line of Talk track for you, like you've, you, you've been a genuine believer for a long time.[00:39:28] swyx: And,[00:39:28] Sam Schillace: and just to be clear, like I haven't been at Microsoft that long. I've only been here for like two, a little over two years and you know, it's a little bit weird for me 'cause for a lot of my career they were the competitor and the enemy and you know, it's kind of funny to be here, but like it's really remarkable.[00:39:40] On Satya[00:39:40] Sam Schillace: It's going on. I really, really like Satya. I've met a, met and worked with a bunch of big tech CEOs and I think he's a genuinely awesome person and he's fun to work with and has a really great. vision. So like, and I obviously really like Kevin, we've been friends for a long time. So it's a cool place.[00:39:56] Sam Schillace: I think there's a lot of interesting stuff. We[00:39:57] swyx: have some awareness Satya is a listener. So obviously he's super welcome on the pod anytime. You can just drop in a good word for us.[00:40:05] Sam Schillace: He's fun to talk to. It's interesting because like CEOs can be lots of different personalities, but he is you were asking me about how I'm like, so hands on and engaged.[00:40:14] Sam Schillace: I'm amazed at how hands on and engaged he can be given the scale of his job. Like, he's super, super engaged with stuff, super in the details, understands a lot of the stuff that's going on. And the science side of things, as well as the product and the business side, I mean, it's really remarkable. I don't say that, like, because he's listening or because I'm trying to pump the company, like, I'm, like, genuinely really, really impressed with, like, how, what he's, like, I look at him, I'm like, I love this stuff, and I spend all my time thinking about it, and I could not do what he's doing.[00:40:42] Sam Schillace: Like, it's just incredible how much you can get[00:40:43] Ben Dunphy: into his head.[00:40:44] Sam at AI Leadership Track[00:40:44] Ben Dunphy: Sam, it's been an absolute pleasure to hear from you here, hear the war stories. So thank you so much for coming on. Quick question though you're here on the podcast as the presenting sponsor for the AI Engineer World's Fair, will you be taking the stage there, or are we going to defer that to Satya?[00:41:01] Ben Dunphy: And I'm happy[00:41:02] Sam Schillace: to talk to folks. I'm happy to be there. It's always fun to like I, I like talking to people more than talking at people. So I don't love giving keynotes. I love giving Q and A's and like engaging with engineers and like. I really am at heart just a builder and an engineer, and like, that's what I'm happiest doing, like being creative and like building things and figuring stuff out.[00:41:22] Sam Schillace: That would be really fun to do, and I'll probably go just to like, hang out with people and hear what they're working on and working about.[00:41:28] swyx: The AI leadership track is just AI leaders, and then it's closed doors, so you know, more sort of an unconference style where people just talk[00:41:34] Sam Schillace: about their issues.[00:41:35] Sam Schillace: Yeah, that would be, that's much more fun. That's really, because we are really all wrestling with this, trying to figure out what it means. Right. So I don't think anyone I, the reason I have the Scalache laws kind of give me the willies a little bit is like, I, I was joking that we should just call them the Scalache best guesses, because like, I don't want people to think that that's like some iron law.[00:41:52] Sam Schillace: We're all trying to figure this stuff out. Right. Like some of it's right. Some it's not right. It's going to be messy. We'll have false starts, but yeah, we're all working it out. So that's the fun conversation. All[00:42:02] Ben Dunphy: right. Thanks for having me. Yeah, thanks so much for coming on.[00:42:05] Final Plug for Tickets & CFP[00:42:05] Ben Dunphy: For those of you listening, interested in attending AI Engineer World's Fair, you can purchase your tickets today.[00:42:11] Ben Dunphy: Learn more about the event at ai. engineer. You can purchase even group discounts. If you purchase four more tickets, use the code GROUP, and one of those four tickets will be free. If you want to speak at the event CFP closes April 8th, so check out the link at ai. engineer, send us your proposals for talks, workshops, or discussion groups.[00:42:33] Ben Dunphy: So if you want to come to THE event of the year for AI engineers, the technical event of the year for AI engineers this is at June 25, 26, and 27 in San Francisco. That's it! Get full access to Latent Space at www.latent.space/subscribe

Guest host Dr. Sarah Bastawrous summarizes the article titled “Multimodality Imaging in Metabolic Syndrome: State-of-the-art Review” from the March 2024 RadioGraphics issue. Multimodality Imaging in Metabolic Syndrome:State-of-the-Art Review. Kalisz et al. RadioGraphics 2024; 44(3):e230083. 

In this week's episode, Katherine Forrest and Anna Gressel take a deeper dive into large language models (LLMs) and multimodal models (MLLMs). What do you need to know about LLMs and MLLMs, and should you buy one or build your own?   ## Learn More About Paul, Weiss's Artificial Intelligence Practice: https://www.paulweiss.com/practices/litigation/artificial-intelligence

The emergence of ChatGPT has sent shockwaves through many secondary and post-secondary English departments. There's no shortage of doomsaying and prognosticating about the future of writing instruction, even the discipline itself, in the wake of the large language model revolution. Luckily for us, my guest today is Dr. J Palmeri—Professor of English and Director of the Writing Program at Georgetown University. J's work exploring the past, present, and future of multimodal composition is some of the richest, most comprehensive scholarship I've seen. Better still, J practices what they preach in the classroom. Over the course of our dialogue, J details the ways they use new media pedagogy to learn with students, embrace play, compose for real audiences, hack technology, center learning, and ultimately to rethink teaching and learning. There is no shortage of philosophical questions and practical suggestions, but my favorite part of this episode is the way J situates his work on multimodality within a broader story—one that will likely resonate with many of you. This episode is a powerful reminder of why technology is only a tool. Whether that technology is tactile, digital, or artificial intelligence, there is no replacing the deeply human parts of teaching, learning, and communicating alongside others.Faculty Page100 Years of New Media Pedagogy (Open Source Book)Academic Research (Google Scholar)Support the show

The Role of Multimodality Imaging for Pericarditis Guest: Prajwal (Praj) Reddy, MD Hosts: Malcolm R. Bell, M.D. Imaging can play a crucial role in diagnosis and management of pericarditis, particularly if recurrent. Echocardiography, cardiac computed tomography (CCT), and cardiac magnetic resonance imaging (CMR) offer complimentary evaluation of pericardial disease in its various presentations. In this podcast, we review the characteristic signs and features on multi-modality imaging in patients with acute and recurrent pericardial inflammation and its utility in tailoring therapy.   Topics Discussed: What is the role of imaging in the initial diagnosis of pericarditis? How is multi-modality cardiac imaging utilized in recurrent pericarditis? Follow up question to the above: Is serial imaging with cardiac MRI helpful in treatment of recurrent pericarditis? Connect with Mayo Clinic's Cardiovascular Continuing Medical Education online at https://cveducation.mayo.edu or on Twitter @MayoClinicCV and @MayoCVservices. LinkedIn: Mayo Clinic Cardiovascular Services Cardiovascular Education App: The Mayo Clinic Cardiovascular CME App is an innovative educational platform that features cardiology-focused continuing medical education wherever and whenever you need it. Use this app to access other free content and browse upcoming courses. Download it for free in Apple or Google stores today! No CME credit offered for this episode. Podcast episode transcript found here.

In this new podcast episode, we dive deep into the realms of somatic psychotherapy, interdisciplinary practices, and the art of embodying multi-modality in therapy with our distinguished guest, Dr. Brian Tierney. Known as the Somatic Doctor, Brian brings a unique blend of expertise in somatic psychotherapy, dance, and bodywork, shining a light on the intricate connections between the body, mind, and spirit.Throughout our conversation, Brian shares his journey from a business student with a budding interest in somatic practices to becoming a leader in the field of integrative therapy. His eclectic approach, combining talk therapy, dance, and structural integration bodywork, offers a fresh perspective on therapeutic practices. Brian's work emphasizes the importance of navigating multiple therapeutic angles, including trauma resolution, dance and movement, and the integration of various bodywork modalities, to support human development and well-being.A highlight of our discussion revolves around Brian's deep commitment to interdisciplinary practice and research. He challenges the trend towards excessive specialization in therapy, advocating for a broader, more holistic approach to understanding and treating the human psyche. Brian's insights into the role of dance, movement, and body awareness in therapy provide valuable lessons on the power of embodying multi-modality to foster healing and personal growth.Listeners are invited to explore the full episode to delve into Brian's profound insights and experiences. For more information on Dr. Brian Tierney's work, visit his website at somaticdoctor.com and follow him on Instagram. Engage with his teachings and join us in exploring the vast potential of integrative and somatic therapies to enrich our lives and therapeutic practices.Key Highlights:00:00 Introduction to Dr. Brian Tierney02:12 Interdisciplinary practice and research02:54 Origin story and path to somatic psychotherapy04:26 The transformative power of bodywork and dance06:23 The journey through graduate studies and beyond08:30 The significance of men's work in therapy11:59 An integrative approach to psychology17:01 The tension between opposites in therapy21:51 The future of multimodal therapeutic practices25:39 Self-regulation and the role of dynamic experiences29:04 Embracing playfulness in therapy33:01 The challenge of othering in therapeutic communities:37:06 Brian's clinical practice and teaching philosophy41:27 The importance of facial work in therapy46:18 Polyvagal theory and its place in modern therapy50:43 The evolving landscape of somatic psychotherapyFor those interested in the intersection of somatic practices, dance, and psychotherapy, this episode is a must-listen. Dive into the full conversation to uncover the depths of Dr. Brian Tierney's work and the transformative potential of integrating multiple modalities into therapeutic practice.Links and Resources Mentioned:  Website: https://somaticdoctor.com/Facebook: https://www.facebook.com/SomaticDoctor/ Instagram: https://www.instagram.com/thesomaticdoctor 

This episode is all about the different modes of communication that we use beyond the 4 linguistic skills of reading, writing, listening speaking. My guest is Nik Peachy who has helped to write a new paper published by OUP called "Multimodality in ELT: Communication Skills for Today's Generation". Listen to Nik and me chatting about the importance of multimodal literacy in our social interactions and in the ways we consume and produce media online.

Note for Latent Space Community members: we have now soft-launched meetups in Singapore, as well as two new virtual paper club/meetups for AI in Action and LLM Paper Club. We're also running Latent Space: Final Frontiers, our second annual demo day hackathon from last year.For the first time, we are doing an audio version of monthly AI Engineering recap that we publish on Latent Space! This month it's “The Four Wars of the AI Stack”; you can find the full recap with all the show notes here: https://latent.space/p/dec-2023* [00:00:00] Intro* [00:01:42] The Four Wars of the AI stack: Data quality, GPU rich vs poor, Multimodality, and Rag/Ops war* [00:03:17] Selection process for the four wars and notable mentions* [00:06:58] The end of low background tokens and the impact on data engineering* [00:08:36] The Quality Data Wars (UGC, licensing, synthetic data, and more)* [00:14:51] Synthetic Data* [00:17:49] The GPU Rich/Poors War* [00:18:21] Anyscale benchmark drama* [00:22:00] The math behind Mixtral inference costs* [00:28:48] Transformer alternatives and why they matter* [00:34:40] The Multimodality Wars* [00:38:10] Multiverse vs Metaverse* [00:45:00] The RAG/Ops Wars* [00:50:00] Will frameworks expand up, or will cloud providers expand down?* [00:54:32] Syntax to Semantics* [00:56:41] Outer Loop vs Inner Loop* [00:59:54] Highlight of the month Get full access to Latent Space at www.latent.space/subscribe

Happy 2024! We appreciated all the feedback on the listener survey (still open, link here)! Surprising to see that some people's favorite episodes were others' least, but we'll always work on improving our audio quality and booking great guests. Help us out by leaving reviews on Twitter, YouTube, and Apple Podcasts!

We're sharing a few of Nathan's favorite AI scouting episodes from other shows. Today, Shane Legg, Cofounder at Deepmind and its current Chief AGI Scientist, shares his insights with Dwarkesh Patel on AGI's timeline, the new architectures needed for AGI, and why multimodality will be the next big landmark. If you need an ecommerce platform, check out our sponsor Shopify: https://shopify.com/cognitive for a $1/month trial period. You can subscribe to The Dwarkesh Podcast here: https://www.youtube.com/@DwarkeshPatel We're hiring across the board at Turpentine and for Erik's personal team on other projects he's incubating. He's hiring a Chief of Staff, EA, Head of Special Projects, Investment Associate, and more. For a list of JDs, check out: eriktorenberg.com. --- SPONSORS: Shopify is the global commerce platform that helps you sell at every stage of your business. Shopify powers 10% of ALL eCommerce in the US. And Shopify's the global force behind Allbirds, Rothy's, and Brooklinen, and 1,000,000s of other entrepreneurs across 175 countries.From their all-in-one e-commerce platform, to their in-person POS system – wherever and whatever you're selling, Shopify's got you covered. With free Shopify Magic, sell more with less effort by whipping up captivating content that converts – from blog posts to product descriptions using AI. Sign up for $1/month trial period: https://shopify.com/cognitive Omneky is an omnichannel creative generation platform that lets you launch hundreds of thousands of ad iterations that actually work customized across all platforms, with a click of a button. Omneky combines generative AI and real-time advertising data. Mention "Cog Rev" for 10% off www.omneky.com NetSuite has 25 years of providing financial software for all your business needs. More than 36,000 businesses have already upgraded to NetSuite by Oracle, gaining visibility and control over their financials, inventory, HR, eCommerce, and more. If you're looking for an ERP platform ✅ head to NetSuite: http://netsuite.com/cognitive and download your own customized KPI checklist. X/SOCIAL: @labenz (Nathan) @dwarkesh_sp (Dwarkesh) @shanelegg (Shane) @CogRev_Podcast (Cognitive Revolution) TIMESTAMPS: (00:00:00) - Episode Preview with Nathan's Intro (00:02:45) - Conversation with Dwarkesh and Shane begins (00:14:26) - Do we need new architectures? (00:17:31) - Sponsors: Shopify  (00:19:40) - Is search needed for creativity? (00:31:46) - Impact of Deepmind on safety vs capabilities (00:32:48) - Sponsors: Netsuite | Omneky (00:37:10) - Timelines (00:45:18) - Multimodality

CardioNerds cofounder Dr. Amit Goyal and cardiology fellows from the Cleveland Clinic (Drs. Alejandro Duran Crane, Gary Parizher, and Simrat Kaur) discuss the following case: A 61-year-old man presented with symptoms of heart failure and left ventricular hypertrophy. He was given a diagnosis of obstructive hypertrophic cardiomyopathy. He eventually underwent septal myectomy, mitral valve replacement, aortic aneurysm repair, and aortic valve replacement with findings of Fabry's disease on surgical pathology. The case discussion focuses on the differential diagnosis for LVH and covers Fabry disease as an HCM mimic. Expert commentary was provided by Dr. Angelika Ewrin. The episode audio was edited by student Dr. Diane Masket. US Cardiology Review is now the official journal of CardioNerds! Submit your manuscript here. CardioNerds Case Reports PageCardioNerds Episode PageCardioNerds AcademyCardionerds Healy Honor Roll CardioNerds Journal ClubSubscribe to The Heartbeat Newsletter!Check out CardioNerds SWAG!Become a CardioNerds Patron! Case Media - An Unusual Cause of Hypertrophic Cardiomyopathy – Cleveland Clinic Pearls - An Unusual Cause of Hypertrophic Cardiomyopathy – Cleveland Clinic Left ventricular hypertrophy is a cardiac manifestation of several different systemic and cardiac processes, and its etiology should be clarified to avoid missed diagnosis and treatment opportunities. Fabry disease is a rare, X-linked inherited disease that can present cardiac and extra-cardiac manifestations, the former of which include hypertrophic cardiomyopathy, conduction defects, coronary artery disease, conduction abnormalities, arrhythmias, and heart failure.  The diagnosis of Fabry disease includes measurement of alpha-galactosidase enzyme activity as well as genetic testing to evaluate for pathogenic variants or variants of unknown significance in the GLA gene. Family members of patients diagnosed with Fabry disease should be screened based on the inheritance pattern.   Multimodality imaging can be helpful in the diagnosis of Fabry disease. Echocardiography can show left ventricular hypertrophy (LVH), reduced global strain, aortic and mitral valve thickening, and aortic root dilation with associated mild to moderate aortic regurgitation. Cardiac MRI can show hypertrophy of papillary muscles, mid-wall late gadolinium enhancement and low-native T1 signal.   The treatment of Fabry disease involves a multi-disciplinary approach with geneticists, nephrologists, cardiologists, nephrologists, and primary care doctors. Enzyme replacement therapy can delay the progression of cardiac disease.    Show Notes - An Unusual Cause of Hypertrophic Cardiomyopathy – Cleveland Clinic What are the causes of left ventricular hypertrophy? LVH is extremely common. It is present in 15-20% of the general population, and is more common in Black individuals, the elderly, obese or hypertensive individuals, with most cases being secondary to hypertension and aortic valve stenosis. In general terms, it is helpful to divide the causes of LVH into three main groups: high afterload states, obstruction to LV ejection, and intrinsic myocardial problems. Increased afterload states include both primary and secondary hypertension and renal artery stenosis. Mechanical obstruction includes aortic stenosis, subaortic stenosis, and coarctation of the aorta. Lastly, several intrinsic problems of the myocardium can cause LV hypertrophy, such as athletic heart with physiological LVH, hypertrophic cardiomyopathy with or without outflow obstruction, and infiltrative or storage diseases such as cardiac amyloidosis, Fabry's disease, or Danon disease, among others.  How does Fabry disease present? Fabry disease is present in all races and is an X-linked lysosomal storage disorder caused by pathogenic variants in the GLA gene that result in reduced alpha-galactosidase enzyme activity,

What new possibilities do you see emerging with voice technology? How might it influence our interactions with businesses and services in the future and what if your voice could transform the way we interact with technology? Imagine a world where your voice can effortlessly interact with devices and transform the way we navigate our surroundings.In this episode of This Anthro Life, we explore the world of future technology with guest Tobias Dengel, a leading expert in digital transformation, and discuss the power of voice technology and its potential to transform how we interact with devices and the world around us. Dengel sheds light on the reasons why integrating voice technology into existing platforms is perceived as a safer approach compared to building entirely new platforms from scratch. He emphasizes the importance of leveraging the familiarity and trust already established with these platforms, enabling a smoother transition for users. Additionally, Dengel delves into the widespread adoption of voice assistants such as Alexa and Siri, highlighting their increasing presence in our daily lives. Furthermore, the discussion extends to the role of voice technology in banking applications, where it plays a crucial role in enhancing security measures and making our lives safer. The exploration of voice technology in this episode showcases its transformative potential and the various ways it is revolutionizing our interactions with devices and services.Tune in to discover Dengel's captivating insights and expertise as we envision the transformative power of voice technology.Key takeaways:Voice technology is evolving and becoming increasingly sophisticated, with the adoption of voice assistants like Alexa and Siri skyrocketing.Adding voice to existing platforms feels safer than creating new ones altogether, as users are already familiar with the platform and trust it.Voice technology solves the problem of faster communication, as humans speak three times faster than they type.The interface of voice technology needs to be redesigned to be more efficient, as listening to machines is slower than reading or interacting with visuals.The more human-like voice assistants become, the less users trust them, as they feel like they are being tricked.Multimodality is important in voice technology, as it allows for a combination of voice, visuals, and other forms of communication to enhance the user experience.Voice technology has applications in various industries, such as law enforcement, warehouses, retail, and safety in industrial settings.The combination of generative AI and conversational AI is where the magic happens in voice technology, allowing for more accurate interpretation and response.Conversational designers will play a crucial role in designing effective voice experiences, considering factors like speed, efficiency, and user preferences.Voice technology has the potential to reshape business processes and models, such as centralized restaurants, telemedicine, and global healthcare access.Timestamps:00:00:07 Voice technology is evolving.00:05:14 Design voice experiences in multimodal.00:09:37 Voice is a powerful interface.00:18:38 Conversational AI and generative AI.00:21:08 Context is crucial for conversation.00:28:31 The blend of generative AI and conversational AI is creating a user experience breakthrough.00:29:15 Voice experiences are becoming multimodal.00:34:06 Voice technology revolutionizes business processes.00:39:03 The future of technology is voice-based.00:43:30 Spread anthropological thinking to audiences.Tobias Dengel is a seasoned technology executive with over 20 years of experience in mobility, digital media, and interactive marketing. He currently holds the position of President at WillowTree, a TELUS International Company, a global leader in digital product design and development. Dengel's expertise and leadership have contributed to WillowTree's continuous growth and recognition as one of America's fastest-growing companies, as listed by Inc. magazine for 11 consecutive years. He is also the author of the book "The Sound of the Future: The Coming Age of AI-Enabled Voice Technology," where he explores the transformative potential of voice technology in various aspects of business and society.About This Anthro Life This Anthro Life is a thought-provoking podcast that explores the human side of technology, culture, and business. Hosted by Adam Gamwell, we unravel fascinating narratives and connect them to the wider context of our lives. Tune in to https://thisanthrolife.org and subscribe to our Substack at https://thisanthrolife.substack.com for more captivating episodes and engaging content.Connect with Tobias DengelLinkedin: https://www.linkedin.com/in/tobiasdengel/ Twitter: https://x.com/TobiasDengel?s=20Website: https://www.tobiasdengel.com/ Facebook: https://www.facebook.com/tobias.denge.7/ Connect with This Anthro Life:Instagram: https://www.instagram.com/thisanthrolife/ Facebook: https://www.facebook.com/thisanthrolife LinkedIn: https://www.linkedin.com/company/this-anthro-life-podcast/ This Anthro Life website: https://www.thisanthrolife.org/ Substack blog: https://thisanthrolife.substack.comThis show is part of the Spreaker Prime Network, if you are interested in advertising on this podcast, contact us at https://www.spreaker.com/show/5168968/advertisement

Guest host Dr. Gagandeep Choudhary summarizes the article “Multimodality Imaging of Dementia: Clinical Importance and Role of Integrated Anatomic and Molecular Imaging” focusing on pathophysiology and imaging findings of some of the major neurodegenerative processes.  Multimodality Imaging of Demen-tia: Clinical Importance and Role of Integrated Anatomic and Mo-lecular Imaging. Patel et al. RadioGraphics 2020; 40:200–222.

In episode 92 of The Gradient Podcast, Daniel Bashir speaks to Kevin K. Yang.Kevin is a senior researcher at Microsoft Research (MSR) who works on problems at the intersection of machine learning and biology, with an emphasis on protein engineering. He completed his PhD at Caltech with Frances Arnold on applying machine learning to protein engineering. Before joining MSR, he was a machine learning scientist at Generate Biomedicines, where he used machine learning to optimize proteins.Have suggestions for future podcast guests (or other feedback)? Let us know here or reach us at editor@thegradient.pubSubscribe to The Gradient Podcast:  Apple Podcasts  | Spotify | Pocket Casts | RSSFollow The Gradient on TwitterOutline:* (00:00) Intro* (02:40) Kevin's background* (06:00) Protein engineering early in Kevin's career* (12:10) From research to real-world proteins: the process* (17:40) Generative models + pretraining for proteins* (22:47) Folding diffusion for protein structure generation* (30:45) Protein evolutionary dynamics and generative models of protein sequences* (40:03) Analogies and disanalogies between protein modeling and language models* (41:45) In representation learning* (45:50) Convolutions vs. transformers and inductive biases* (49:25) Pretraining tasks for protein structure* (51:45) More on representation learning for protein structure* (54:06) Kevin's thoughts on interpretability in deep learning for protein engineering* (56:50) Multimodality in protein engineering and future directions* (59:14) OutroLinks:* Kevin's Twitter and homepage* Research* Generative models + pre-training for proteins and chemistry* Broad intro to techniques in the space* Protein structure generation via folding diffusion* Protein sequence design with deep generative models (review)* Evolutionary velocity with protein language models predicts evolutionary dynamics of diverse proteins* Protein generation with evolutionary diffusion: sequence is all you need* ML for protein engineering* ML-guided directed evolution for protein engineering (review)* Learned protein embeddings for ML* Adaptive machine learning for protein engineering (review)* Multimodal deep learning for protein engineering Get full access to The Gradient at thegradientpub.substack.com/subscribe

#mentalhealth #neurofeedback #biofeedback #photobiomodulation #hrv #oxygentherapy #rebeccabassham #jaygunkelman #mentalhealthpodcast #mentalhealthshorts Main Show: https://youtu.be/kZXUrEssxUQ Rebecca Bassham BS, RT, BCN, QEEG-T Owner Edge NeuroFitness http://www.edgeneurofitness.com/ joins Jay Gunkelman the man who has read well over 500,000 brain scans, Dr. Mari Swingle the author of iMinds http://www.drmariswingle.com/about-dr-mari/ and Pete Jansons on the NeuroNoodle Neurofeedback Podcast to give us an update on what she has been working on since we last saw her at the Suisun City Neurofeedback Summit in 2022. In this short clip Rebecca Bassham demonstrates the multi modal approach to mental health with Neurofeedback Therapy, Biofeedback Therapy, HRV, Photobiomodulation and Oxygen Therapy There are seats left to this years Suisun Summit https://greymattersneurofeedback.com/suisun-summit/ --- Send in a voice message: https://podcasters.spotify.com/pod/show/neuronoodle/message Support this podcast: https://podcasters.spotify.com/pod/show/neuronoodle/support

Video dominates modern media consumption, but video creation is still expensive and difficult. AI-generated and edited video is a holy grail of democratized creative expression. This week on No Priors, Sarah Guo and Elad Gil sit down with Devi Parikh. She is a Research Director in Generative AI at Meta and an Associate Professor in the School of Interactive Computing at Georgia Tech. Her work focuses on multimodality and AI for images, audio and video. Recently, she worked on Make a Video 3D, also called MAV3D, which creates animations from text prompts. She is also a talented AI-generated and analog artist herself. Elad, Sarah and Devi talk about what's exciting in computer vision, what's blocking researchers from fully immersive Generative 4-D, and AI controllability. No Priors is now on YouTube! Subscribe to the channel on YouTube and like this episode. Show Links: Devi Parikh - Google Scholar  Text-To-4D Dynamic Scene Generation named MAV3D (Make-A-Video3D) Full Research Paper Website with examples of image to 4 D generation Devi's Substack Sign up for new podcasts every week. Email feedback to show@no-priors.com Follow us on Twitter: @NoPriorsPod | @Saranormous | @EladGil | @DeviParikh Show Notes: (0:00:06) - Democratizing Creative Expression With AI-Generated Video (0:08:31) - Challenges in Video Generation Research (0:15:57) - Challenges and Implications of Video Processing (0:20:43) - Control and Multi-Modal Inputs in Video (0:25:50) - Audio's Role in Visual Content (0:39:00) - Don't Self-Select & Devi's tips for young researchers

In this episode, CardioNerds co-founder Amit Goyal joins Dr. Iva Minga, Dr. Kevin Lee, and Dr. Juan Pablo Salazar Adum from the University of Chicago - Northshore in Evanston, IL to discuss a case of primary cardiac diffuse large B-cell lymphoma. The ECPR for this episode is provided by Dr. Amit Pursnani (Advanced Cardiac Imaging, Fellowship program director, NorthShore University HealthSystem). Audio editing by CardioNerds Academy Intern, Dr. Akiva Rosenzveig. Case synopsis: A 77-year-old man with no significant medical history presents to the emergency department with progressive shortness of breath for 1 week. He reports an unintentional 15-pound weight loss in the prior month as well as constipation and abdominal/flank pain. On examination he was found to be tachycardic with a regular rhythm and further evaluation with a chest X-ray and chest CT scan demonstrated a large pericardial effusion. This was further investigated with an urgent echocardiogram that revealed a large pericardial effusion with a large mass attached to the pericardial side of the RV free wall, as well as signs of early cardiac tamponade. A pericardiocentesis was performed and 550mL of bloody fluid was withdrawn. The fluid was sent for laboratory analysis and cytology. A cardiac MRI demonstrated a large invasive mass in the pericardium and RV wall consistent with cardiac lymphoma. Cytology confirmed diffuse large B-cell lymphoma. Subsequent CT and PET scans did not find any other site of malignancy, giving the patient a diagnosis of primary cardiac diffuse large B-cell lymphoma. The patient underwent R-CHOP chemotherapy and was followed closely with repeat cardiac MRI and PET scans which demonstrated resolution of the cardiac mass at his one-year surveillance follow-up. This case was published in US Cardiology Review, the official journal of CardioNerds. To learn more, access the case report article here. CardioNerds is collaborating with Radcliffe Cardiology and US Cardiology Review journal (USC) for a ‘call for cases', with the intention to co-publish high impact cardiovascular case reports, subject to double-blind peer review. Case Reports that are accepted in USC journal and published as the version of record (VOR), will also be indexed in Scopus and the Directory of Open Access Journals (DOAJ). CardioNerds Case Reports PageCardioNerds Episode PageCardioNerds AcademyCardionerds Healy Honor Roll CardioNerds Journal ClubSubscribe to The Heartbeat Newsletter!Check out CardioNerds SWAG!Become a CardioNerds Patron! Pearls - A Mystery Mass in the Heart - Cardiac Lymphoma The most common cause of malignant cardiac masses is metastasis. Primary cardiac tumors are rare. Cardiac tumors are separated into 2 categories: benign and malignant. They are often differentiated based on their location and their degree of tissue invasion. Multimodality imaging is essential in the diagnosis, management, and surveillance of cardiac masses. A multidisciplinary team approach is invaluable for management of patients with cardiac tumors. Show Notes - A Mystery Mass in the Heart - Cardiac Lymphoma 1. What is the clinical presentation of cardiac masses? Cardiac masses can have a variable presentation. They can present with arrhythmias, angina, heart failure symptoms, or pericardial effusion. Patients can also be asymptomatic; the masses can be found incidentally on cardiac or chest imagining. 2. What is the differential diagnosis for cardiac masses? Cardiac masses are separated into benign and malignant. The most common malignant cardiac masses are metastases from a distant source. The location of the mass is important in narrowing the differential. 3. What imaging modalities are used to diagnose cardiac masses? Multimodality imaging is needed to describe the mass in detail and guide diagnosis. An echocardiogram is usually the first imaging modality. Cardiac MRI is a great modality that allows for the...

Thanks to the over 42,000 latent space explorers who checked out our Replit episode! We are hosting/attending a couple more events in SF and NYC this month. See you if in town!Lexica.art was introduced to the world 24 hours after the release of Stable Diffusion as a search engine for prompts, gaining instant product-market fit as a world discovering generative AI also found they needed to learn prompting by example.Lexica is now 8 months old, serving 5B image searches/day, and just shipped V3 of Lexica Aperture, their own text-to-image model! Sharif Shameem breaks his podcast hiatus with us for an exclusive interview covering his journey building everything with AI!The conversation is nominally about Sharif's journey through his three startups VectorDash, Debuild, and now Lexica, but really a deeper introspection into what it takes to be a top founder in the fastest moving tech startup scene (possibly ever) of AI. We hope you enjoy this conversation as much as we did!Full transcript is below the fold. We would really appreciate if you shared our pod with friends on Twitter, LinkedIn, Mastodon, Bluesky, or your social media poison of choice!Timestamps* [00:00] Introducing Sharif* [02:00] VectorDash* [05:00] The GPT3 Moment and Building Debuild* [09:00] Stable Diffusion and Lexica* [11:00] Lexica's Launch & How it Works* [15:00] Being Chronically Early* [16:00] From Search to Custom Models* [17:00] AI Grant Learnings* [19:30] The Text to Image Illuminati?* [20:30] How to Learn to Train Models* [24:00] The future of Agents and Human Intervention* [29:30] GPT4 and Multimodality* [33:30] Sharif's Startup Manual* [38:30] Lexica Aperture V1/2/3* [40:00] Request for AI Startup - LLM Tools* [41:00] Sequencing your Genome* [42:00] Believe in Doing Great Things* [44:30] Lightning RoundShow Notes* Sharif's website, Twitter, LinkedIn* VectorDash (5x cheaper than AWS)* Debuild Insider, Fast company, MIT review, tweet, tweet* Lexica* Introducing Lexica* Lexica Stats* Aug: “God mode” search* Sep: Lexica API * Sept: Search engine with CLIP * Sept: Reverse image search* Nov: teasing Aperture* Dec: Aperture v1* Dec - Aperture v2* Jan 2023 - Outpainting* Apr 2023 - Aperture v3* Same.energy* AI Grant* Sharif on Agents: prescient Airpods tweet, Reflection* MiniGPT4 - Sharif on Multimodality* Sharif Startup Manual* Sharif Future* 23andMe Genome Sequencing Tool: Promethease* Lightning Round* Fave AI Product: Cursor.so. Swyx ChatGPT Menubar App.* Acceleration: Multimodality of GPT4. Animated Drawings* Request for Startup: Tools for LLMs, Brex for GPT Agents* Message: Build Weird Ideas!TranscriptAlessio: Hey everyone, welcome to the Latent Space podcast. This is Alessio, partner and CTO on Residence at Decibel Partners. I'm joined by my co-host Wix, writer and editor of Latent Space. And today we have Sharish Amin. Welcome to the studio. Sharif: Awesome. Thanks for the invite.Swyx: Really glad to have you. [00:00] Introducing SharifSwyx: You've been a dream guest, actually, since we started drafting guest lists for this pod. So glad we could finally make this happen. So what I like to do is usually introduce people, offer their LinkedIn, and then prompt you for what's not on your LinkedIn. And to get a little bit of the person behind the awesome projects. So you graduated University of Maryland in CS. Sharif: So I actually didn't graduate, but I did study. Swyx: You did not graduate. You dropped out. Sharif: I did drop out. Swyx: What was the decision behind dropping out? Sharif: So first of all, I wasn't doing too well in any of my classes. I was working on a side project that took up most of my time. Then I spoke to this guy who ended up being one of our investors. And he was like, actually, I ended up dropping out. I did YC. And my company didn't end up working out. And I returned to school and graduated along with my friends. I was like, oh, it's actually a reversible decision. And that was like that. And then I read this book called The Case Against Education by Brian Kaplan. So those two things kind of sealed the deal for me on dropping out. Swyx: Are you still on hiatus? Could you still theoretically go back? Sharif: Theoretically, probably. Yeah. Still on indefinite leave. Swyx: Then you did some work at Mitra? Sharif: Mitra, yeah. So they're lesser known. So they're technically like an FFRDC, a federally funded research and development center. So they're kind of like a large government contractor, but nonprofit. Yeah, I did some computer vision work there as well. [02:00] VectorDashSwyx: But it seems like you always have an independent founder bone in you. Because then you started working on VectorDash, which is distributed GPUs. Sharif: Yes. Yeah. So VectorDash was a really fun project that we ended up working on for a while. So while I was at Mitra, I had a friend who was mining Ethereum. This was, I think, 2016 or 2017. Oh my God. Yeah. And he was mining on his NVIDIA 1080Ti, making around like five or six dollars a day. And I was trying to train a character recurrent neural network, like a character RNN on my iMessage text messages to make it like a chatbot. Because I was just curious if I could do it. Because iMessage stores all your past messages from years ago in a SQL database, which is pretty nifty. But I wanted to train it. And I needed a GPU. And it was, I think, $60 to $80 for a T4 on AWS, which is really slow compared to a 1080Ti. If you normalize the cost and performance versus the 1080Ti when someone's mining Ethereum, it's like a 20x difference. So I was like, hey, his name was Alex. Alex, I'll give you like 10 bucks if you let me borrow your 1080Ti for a week. I'll give you 10 bucks per day. And it was like 70 bucks. And I used it to train my model. And it worked great. The model was really bad, but the whole trade worked really great. I got a really high performance GPU to train my model on. He got much more than he was making by mining Ethereum. So we had this idea. I was like, hey, what if we built this marketplace where people could rent their GPUs where they're mining cryptocurrency and machine learning researchers could just rent them out and pay a lot cheaper than they would pay AWS. And it worked pretty well. We launched in a few months. We had over 120,000 NVIDIA GPUs on the platform. And then we were the cheapest GPU cloud provider for like a solid year or so. You could rent a pretty solid GPU for like 20 cents an hour. And cryptocurrency miners were making more than they would make mining crypto because this was after the Ethereum crash. And yeah, it was pretty cool. It just turns out that a lot of our customers were college students and researchers who didn't have much money. And they weren't necessarily the best customers to have as a business. Startups had a ton of credits and larger companies were like, actually, we don't really trust you with our data, which makes sense. Yeah, we ended up pivoting that to becoming a cloud GPU provider for video games. So we would stream games from our GPUs. Oftentimes, like many were located just a few blocks away from you because we had the lowest latency of any cloud GPU provider, even lower than like AWS and sometimes Cloudflare. And we decided to build a cloud gaming platform where you could pretty much play your own games on the GPU and then stream it back to your Mac or PC. Swyx: So Stadia before Stadia. Sharif: Yeah, Stadia before Stadia. It's like a year or so before Stadia. Swtx: Wow. Weren't you jealous of, I mean, I don't know, it sounds like Stadia could have bought you or Google could have bought you for Stadia and that never happened? Sharif: It never happened. Yeah, it didn't end up working out for a few reasons. The biggest thing was internet bandwidth. So a lot of the hosts, the GPU hosts had lots of GPUs, but average upload bandwidth in the United States is only 35 megabits per second, I think. And like a 4K stream needs like a minimum of 15 to 20 megabits per second. So you could really only utilize one of those GPUs, even if they had like 60 or 100. [05:00] The GPT3 Moment and Building DebuildSwyx: And then you went to debuild July 2020, is the date that I have. I'm actually kind of just curious, like what was your GPT-3 aha moment? When were you like GPT-3-pilled? Sharif: Okay, so I first heard about it because I was also working on another chatbot. So this was like after, like everything ties back to this chatbot I'm trying to make. This was after working on VectorDash. I was just like hacking on random projects. I wanted to make the chatbot using not really GPT-2, but rather just like it would be pre-programmed. It was pretty much you would give it a goal and then it would ask you throughout the week how much progress you're making to that goal. So take your unstructured response, usually a reply to a text message, and then it would like, plot it for you in like a table and you could see your progress over time. It could be for running or tracking calories. But I wanted to use GPT-3 to make it seem more natural because I remember someone on Bookface, which is still YC's internal forum. They posted and they were like, OpenAI just released AGI and it's GPT-3. I asked it like a bunch of logic puzzles and it solved them all perfectly. And I was like, what? How's no one else talking about this? Like this is either like the greatest thing ever that everyone is missing or like it's not that good. So like I tweeted out if anyone could get me access to it. A few hours later, Greg Brockman responded. Swyx: He is everywhere. Sharif: He's great. Yeah, he's on top of things. And yeah, by that afternoon, I was like messing around with the API and I was like, wow, this is incredible. You could chat with fake people or people that have passed away. You could like, I remember the first conversation I did was this is a chat with Steve Jobs and it was like, interviewer, hi. What are you up to today on Steve? And then like you could talk to Steve Jobs and it was somewhat plausible. Oh, the thing that really blew my mind was I tried to generate code with it. So I'd write the function for a JavaScript header or the header for a JavaScript function. And it would complete the rest of the function. I was like, whoa, does this code actually work? Like I copied it and ran it and it worked. And I tried it again. I gave more complex things and like I kind of understood where it would break, which was like if it was like something, like if it was something you couldn't easily describe in a sentence and like contain all the logic for in a single sentence. So I wanted to build a way where I could visually test whether these functions were actually working. And what I was doing was like I was generating the code in the playground, copying it into my VS code editor, running it and then reloading the react development page. And I was like, okay, cool. That works. So I was like, wait, let me just put this all in like the same page so I can just compile in the browser, run it in the browser and then submit it to the API in the browser as well. So I did that. And it was really just like a simple loop where you just type in the prompt. It would generate the code and then compile it directly in the browser. And it showed you the response. And I did this for like very basic JSX react components. I mean, it worked. It was pretty mind blowing. I remember staying up all night, like working on it. And it was like the coolest thing I'd ever worked on at the time so far. Yeah. And then I was like so mind blowing that no one was talking about this whole GPT three thing. I was like, why is this not on everyone's minds? So I recorded a quick 30 second demo and I posted on Twitter and like I go to bed after staying awake for like 20 hours straight. When I wake up the next morning and I had like 20,000 likes and like 100,000 people had viewed it. I was like, oh, this is so cool. And then I just kept putting demos out for like the next week. And yeah, that was like my GPT three spark moment. Swyx: And you got featured in like Fast Company, MIT Tech Review, you know, a bunch of stuff, right? Sharif: Yeah. Yeah. I think a lot of it was just like the API had been there for like a month prior already. Swyx: Not everyone had access. Sharif: That's true. Not everyone had access. Swyx: So you just had the gumption to tweet it out. And obviously, Greg, you know, on top of things as always. Sharif: Yeah. Yeah. I think it also makes a lot of sense when you kind of share things in a way that's easily consumable for people to understand. Whereas if you had shown a terminal screenshot of a generating code, that'd be pretty compelling. But whereas seeing it get rendered and compiled directly in front of you, there's a lot more interesting. There's also that human aspect to it where you want to relate things to the end user, not just like no one really cares about evals. When you can create a much more compelling demo explaining how it does on certain tasks. [09:00] Stable Diffusion and LexicaSwyx: Okay. We'll round it out soon. But in 2022, you moved from Debuild to Lexica, which was the search engine. I assume this was inspired by stable diffusion, but I can get the history there a little bit. Sharif: Yeah. So I was still working on Debuild. We were growing at like a modest pace and I was in the stable... Swyx: I was on the signup list. I never got off. Sharif: Oh yeah. Well, we'll get you off. It's not getting many updates anymore, but yeah, I was in the stable diffusion discord and I was in it for like many hours a day. It was just like the most exciting thing I'd ever done in a discord. It was so cool. Like people were generating so many images, but I didn't really know how to write prompts and people were like writing really complicated things. They would be like, like a modern home training on our station by Greg Rutkowski, like a 4k Unreal Engine. It's like that there's no way that actually makes the images look better. But everyone was just kind of copying everyone else's prompts and like changing like the first few words. Swyx: Yeah. Yeah. Sharif: So I was like using the discord search bar and it was really bad because it showed like five images at a time. And I was like, you know what? I could build a much better interface for this. So I ended up scraping the entire discord. It was like 10 million images. I put them in a database and I just pretty much built a very basic search engine where you could just type for type a word and then it returned all the prompts that had that word. And I built the entire website for it in like 20, in like about two days. And we shipped it the day I shipped it the day after the stable diffusion weights were open sourced. So about 24 hours later and it kind of took off in a way that I never would have expected. Like I thought it'd be this cool utility that like hardcore stable diffusion users would find useful. But it turns out that almost anyone who mentioned stable diffusion would also kind of mention Lexica in conjunction with it. I think it's because it was like it captured the zeitgeist in an easy to share way where it's like this URL and there's this gallery and you can search. Whereas running the model locally was a lot harder. You'd have to like to deploy it on your own GPU and like set up your own environment and like do all that stuff. Swyx: Oh, my takeaway. I have two more to add to the reasons why Lexica works at the time. One is lower latency is all you need. So in other words, instead of waiting a minute for your image, you could just search and find stuff that other people have done. That's good. And then two is everyone knew how to search already, but people didn't know how to prompt. So you were the bridge. Sharif: That's true. Yeah. You would get a lot better looking images by typing a one word prompt versus prompting for that one word. Yeah. Swyx: Yeah. That is interesting. [11:00] Lexica's Explosion at LaunchAlessio: The numbers kind of speak for themselves, right? Like 24 hours post launch, 51,000 queries, like 2.2 terabytes in bandwidth. Going back to the bandwidth problem that you have before, like you would have definitely run into that. Day two, you doubled that. It's like 111,000 queries, four and a half terabytes in bandwidth, 22 million images served. So it's pretty crazy. Sharif: Yeah. I think we're, we're doing like over 5 billion images served per month now. It's like, yeah, that's, it's pretty crazy how much things have changed since then. Swyx: Yeah. I'm still showing people like today, even today, you know, it's been a few months now. This is where you start to learn image prompting because they don't know. Sharif: Yeah, it is interesting. And I, it's weird because I didn't really think it would be a company. I thought it would just be like a cool utility or like a cool tool that I would use for myself. And I really was just building it for myself just because I didn't want to use the Discord search bar. But yeah, it was interesting that a lot of other people found it pretty useful as well. [11:00] How Lexica WorksSwyx: So there's a lot of things that you release in a short amount of time. The God mode search was kind of like, obviously the first thing, I guess, like maybe to talk about some of the underlying technology you're using clip to kind of find, you know, go from image to like description and then let people search it. Maybe talk a little bit about what it takes to actually make the search magic happen. Sharif: Yeah. So the original search was just using Postgres' full text search and it would only search the text contents of the prompt. But I was inspired by another website called Same Energy, where like a visual search engine. It's really cool. Do you know what happened to that guy? I don't. Swyx: He released it and then he disappeared from the internet. Sharif: I don't know what happened to him, but I'm sure he's working on something really cool. He also worked on like Tabnine, which was like the very first version of Copilot or like even before Copilot was Copilot. But yeah, inspired by that, I thought like being able to search images by their semantics. The contents of the image was really interesting. So I pretty much decided to create a search index on the clip embeddings, the clip image embeddings of all the images. And when you would search it, we would just do KNN search on pretty much the image embedding index. I mean, we had way too many embeddings to store on like a regular database. So we had to end up using FAISS, which is a Facebook library for really fast KNN search and embedding search. That was pretty fun to set up. It actually runs only on CPUs, which is really cool. It's super efficient. You compute the embeddings on GPUs, but like you can serve it all on like an eight core server and it's really, really fast. Once we released the semantic search on the clip embeddings, people were using the search way more. And you could do other cool things. You could do like similar image search where if you found like a specific image you liked, you could upload it and it would show you relevant images as well. Swyx: And then right after that, you raised your seed money from AI grant, NetFreedman, then Gross. Sharif: Yeah, we raised about $5 million from Daniel Gross. And then we also participated in AI grant. That was pretty cool. That was kind of the inflection point. Not much before that point, Lexic was kind of still a side project. And I told myself that I would focus on it full time or I'd consider focusing on it full time if we had broke like a million users. I was like, oh, that's gonna be like years away for sure. And then we ended up doing that in like the first week and a half. I was like, okay, there's something here. And it was kind of that like deal was like growing like pretty slowly and like pretty linearly. And then Lexica was just like this thing that just kept going up and up and up. And I was so confused. I was like, man, people really like looking at pictures. This is crazy. Yeah. And then we decided to pivot the entire company and just focus on Lexica full time at that point. And then we raised our seed round. [15:00] Being Chronically EarlySwyx: Yeah. So one thing that you casually dropped out, the one that slip, you said you were working on Lexica before the launch of Stable Diffusion such that you were able to launch Lexica one day after Stable Diffusion. Sharif: Yeah.Swyx: How did you get so early into Stable Diffusion? Cause I didn't hear about it. Sharif: Oh, that's a good question. I, where did I first hear about Stable Diffusion? I'm not entirely sure. It must've been like somewhere on Twitter or something. That changed your life. Yeah, it was great. And I got into the discord cause I'd used Dolly too before, but, um, there were a lot of restrictions in place where you can generate human faces at the time. You can do that now. But when I first got access to it, like you couldn't do any faces. It was like, there were like a, the list of adjectives you couldn't use was quite long. Like I had a friend from Pakistan and it can generate anything with the word Pakistan in it for some reason. But Stable Diffusion was like kind of the exact opposite where there were like very, very few rules. So that was really, really fun and interesting, especially seeing the chaos of like a bunch of other people also using it right in front of you. That was just so much fun. And I just wanted to do something with it. I thought it was honestly really fun. Swyx: Oh, well, I was just trying to get tips on how to be early on things. Cause you're pretty consistently early to things, right? You were Stadia before Stadia. Um, and then obviously you were on. Sharif: Well, Stadia is kind of shut down now. So I don't know if being early to that was a good one. Swyx: Um, I think like, you know, just being consistently early to things that, uh, you know, have a lot of potential, like one of them is going to work out and you know, then that's how you got Lexica. [16:00] From Search to Custom ModelsAlessio: How did you decide to go from search to running your own models for a generation? Sharif: That's a good question. So we kind of realized that the way people were using Lexica was they would have Lexica open in one tab and then in another tab, they'd have a Stable Diffusion interface. It would be like either a discord or like a local run interface, like the automatic radio UI, um, or something else. I just, I would watch people use it and they would like all tabs back and forth between Lexica and their other UI. And they would like to scroll through Lexica, click on the prompt, click on an image, copy the prompt, and then paste it and maybe change a word or two. And I was like, this should really kind of just be all within Lexica. Like, it'd be so cool if you could just click a button in Lexica and get an editor and generate your images. And I found myself also doing the all tab thing, or it was really frustrating. I was like, man, this is kind of tedious. Like I really wish it was much simpler. So we just built generations directly within Lexica. Um, so we do, we deployed it on, I don't remember when we first launched, I think it was November, December. And yeah, people love generating directly within it. [17:00] AI Grant LearningsSwyx: I was also thinking that this was coming out of AI grants where, you know, I think, um, yeah, I was like a very special program. I was just wondering if you learned anything from, you know, that special week where everyone was in town. Sharif: Yeah, that was a great week. I loved it. Swyx: Yeah. Bring us, bring us in a little bit. Cause it was awesome. There. Sharif: Oh, sure. Yeah. It's really, really cool. Like all the founders in AI grants are like fantastic people. And so I think the main takeaway from the AI grant was like, you have this massive overhang in compute or in capabilities in terms of like these latest AI models, but to the average person, there's really not that many products that are that cool or useful to them. Like the latest one that has hit the zeitgeist was chat GPT, which used arguably the same GPT three model, but like RLHF, but you could have arguably built like a decent chat GPT product just using the original GPT three model. But no one really did it. Now there were some restrictions in place and opening. I like to slowly release them over the few months or years after they release the original API. But the core premise behind AI grants is that there are way more capabilities than there are products. So focus on building really compelling products and get people to use them. And like to focus less on things like hitting state of the art on evals and more on getting users to use something. Swyx: Make something people want.Sharif: Exactly. Host: Yeah, we did an episode on LLM benchmarks and we kind of talked about how the benchmarks kind of constrain what people work on, because if your model is not going to do well, unlike the well-known benchmarks, it's not going to get as much interest and like funding. So going at it from a product lens is cool. [19:30] The Text to Image Illuminati?Swyx: My hypothesis when I was seeing the sequence of events for AI grants and then for Lexica Aperture was that you had some kind of magical dinner with Emad and David Holtz. And then they taught you the secrets of training your own model. Is that how it happens? Sharif: No, there's no secret dinner. The Illuminati of text to image. We did not have a meeting. I mean, even if we did, I wouldn't tell you. But it really boils down to just having good data. If you think about diffusion models, really the only thing they do is learn a distribution of data. So if you have high quality data, learn that high quality distribution. Or if you have low quality data, it will learn to generate images that look like they're from that distribution. So really it boils down to the data and the amount of data you have and that quality of that data, which means a lot of the work in training high quality models, at least diffusion models, is not really in the model architecture, but rather just filtering the data in a way that makes sense. So for Lexica, we do a lot of aesthetic scoring on images and we use the rankings we get from our website because we get tens of millions of people visiting it every month. So we can capture a lot of rankings. Oh, this person liked this image when they saw this one right next to it. Therefore, they probably preferred this one over that. You can do pairwise ranking to rank images and then compute like ELO scores. You can also just train aesthetic models to learn to classify a model, whether or not someone will like it or whether or not it's like, rank it on a scale of like one to ten, for example. So we mostly use a lot of the traffic we get from Lexica and use that to kind of filter our data sets and use that to train better aesthetic models. [20:30] How to Learn to Train ModelsSwyx: You had been a machine learning engineer before. You've been more of an infrastructure guy. To build, you were more of a prompt engineer with a bit of web design. This was the first time that you were basically training your own model. What was the wrap up like? You know, not to give away any secret sauce, but I think a lot of people who are traditional software engineers are feeling a lot of, I don't know, fear when encountering these kinds of domains. Sharif: Yeah, I think it makes a lot of sense. And to be fair, I didn't have much experience training massive models at this scale before I did it. A lot of times it's really just like, in the same way when you're first learning to program, you would just take the problem you're having, Google it, and go through the stack overflow post. And then you figure it out, but ultimately you will get to the answer. It might take you a lot longer than someone who's experienced, but I think there are enough resources out there where it's possible to learn how to do these things. Either just reading through GitHub issues for relevant models. Swyx: Oh God. Sharif: Yeah. It's really just like, you might be slower, but it's definitely still possible. And there are really great courses out there. The Fast AI course is fantastic. There's the deep learning book, which is great for fundamentals. And then Andrej Karpathy's online courses are also excellent, especially for language modeling. You might be a bit slower for the first few months, but ultimately I think if you have the programming skills, you'll catch up pretty quickly. It's not like this magical dark science that only three people in the world know how to do well. Probably was like 10 years ago, but now it's becoming much more open. You have open source collectives like Eleuther and LAION, where they like to share the details of their large scale training runs. So you can learn from a lot of those people. Swyx: Yeah. I think what is different for programmers is having to estimate significant costs upfront before they hit run. Because it's not a thing that you normally consider when you're coding, but yeah, like burning through your credits is a fear that people have. Sharif: Yeah, that does make sense. In that case, like fine tuning larger models gets you really, really far. Even using things like low rank adaptation to fine tune, where you can like fine tune much more efficiently on a single GPU. Yeah, I think people are underestimating how far you can really get just using open source models. I mean, before Lexica, I was working on Debuild and we were using the GP3 API, but I was also like really impressed at how well you could get open source models to run by just like using the API, collecting enough samples from like real world user feedback or real world user data using your product. And then just fine tuning the smaller open source models on those examples. And now you have a model that's pretty much state of the art for your specific domain. Whereas the runtime cost is like 10 times or even 100 times cheaper than using an API. Swyx: And was that like GPT-J or are you talking BERT? Sharif: I remember we tried GPT-J, but I think FLAN-T5 was like the best model we were able to use for that use case. FLAN-T5 is awesome. If you can, like if your prompt is small enough, it's pretty great. And I'm sure there are much better open source models now. Like Vicuna, which is like the GPT-4 variant of like Lama fine tuned on like GPT-4 outputs. Yeah, they're just going to get better and they're going to get better much, much faster. Swyx: Yeah. We're just talking in a previous episode to the creator of Dolly, Mike Conover, which is actually commercially usable instead of Vicuna, which is a research project. Sharif: Oh, wow. Yeah, that's pretty cool. [24:00] Why No Agents?Alessio: I know you mentioned being early. Obviously, agents are one of the hot things here. In 2021, you had this, please buy me AirPods, like a demo that you tweeted with the GPT-3 API. Obviously, one of the things about being early in this space, you can only do one thing at a time, right? And you had one tweet recently where you said you hoped that that demo would open Pandora's box for a bunch of weird GPT agents. But all we got were docs powered by GPT. Can you maybe talk a little bit about, you know, things that you wish you would see or, you know, in the last few, last few weeks, we've had, you know, Hugging GPT, Baby AGI, Auto GPT, all these different kind of like agent projects that maybe now are getting closer to the, what did you say, 50% of internet traffic being skips of GPT agents. What are you most excited about, about these projects and what's coming? Sharif: Yeah, so we wanted a way for users to be able to paste in a link for the documentation page for a specific API, and then describe how to call that API. And then the way we would need to pretty much do that for Debuild was we wondered if we could get an agent to browse the docs page, read through it, summarize it, and then maybe even do things like create an API key and register it for that user. To do that, we needed a way for the agent to read the web page and interact with it. So I spent about a day working on that demo where we just took the web page, serialized it into a more compact form that fit within the 2048 token limit of like GPT-3 at the time. And then just decide what action to do. And then it would, if the page was too long, it would break it down into chunks. And then you would have like a sub prompt, decide on which chunk had the best action. And then at the top node, you would just pretty much take that action and then run it in a loop. It was really, really expensive. I think that one 60 second demo cost like a hundred bucks or something, but it was wildly impractical. But you could clearly see that agents were going to be a thing, especially ones that could read and write and take actions on the internet. It was just prohibitively expensive at the time. And the context limit was way too small. But yeah, I think it seems like a lot of people are taking it more seriously now, mostly because GPT-4 is way more capable. The context limit's like four times larger at 8,000 tokens, soon 32,000. And I think the only problem that's left to solve is finding a really good representation for a webpage that allows it to be consumed by a text only model. So some examples are like, you could just take all the text and pass it in, but that's probably too long. You could take all the interactive only elements like buttons and inputs, but then you miss a lot of the relevant context. There are some interesting examples, which I really like is you could run the webpage or you could run the browser in a terminal based browser. So there are some browsers that run in your terminal, which serialize everything into text. And what you can do is just take that frame from that terminal based browser and pass that directly to the model. And it's like a really, really good representation of the webpage because they do things where for graphical elements, they kind of render it using ASCII blocks. But for text, they render it as actual text. So you could just remove all the weird graphical elements, just keep all the text. And that works surprisingly well. And then there are other problems to solve, which is how do you get the model to take an action? So for example, if you have a booking page and there's like a calendar and there are 30 days on the calendar, how do you get it to specify which button to press? It could say 30, and you can match string based and like find the 30. But for example, what if it's like a list of friends in Facebook and trying to delete a friend? There might be like 30 delete buttons. How do you specify which one to click on? The model might say like, oh, click on the one for like Mark. But then you'd have to figure out the delete button in relation to Mark. And there are some ways to solve this. One is there's a cool Chrome extension called Vimium, which lets you use Vim in your Chrome browser. And what you do is you can press F and over every interactive element, it gives you like a character or two characters. Or if you type those two characters, it presses that button or it opens or focuses on that input. So you could combine a lot of these ideas and then get a really good representation of the web browser in text, and then also give the model a really, really good way to control the browser as well. And I think those two are the core part of the problem. The reasoning ability is definitely there. If a model can score in the top 10% on the bar exam, it can definitely browse a web page. It's really just how do you represent text to the model and how do you get the model to perform actions back on the web page? Really, it's just an engineering problem. Swyx: I have one doubt, which I'd love your thoughts on. How do you get the model to pause when it doesn't have enough information and ask you for additional information because you under specified your original request? Sharif: This is interesting. I think the only way to do this is to have a corpus where your training data is like these sessions of agents browsing the web. And you have to pretty much figure out where the ones that went wrong or the agents that went wrong, or did they go wrong and just replace it with, hey, I need some help. And then if you were to fine tune a larger model on that data set, you would pretty much get them to say, hey, I need help on the instances where they didn't know what to do next. Or if you're using a closed source model like GPT-4, you could probably tell it if you're uncertain about what to do next, ask the user for help. And it probably would be pretty good at that. I've had to write a lot of integration tests in my engineering days and like the dome. Alessio: They might be over. Yeah, I hope so. I hope so. I don't want to, I don't want to deal with that anymore. I, yeah, I don't want to write them the old way. Yeah. But I'm just thinking like, you know, we had the robots, the TXT for like crawlers. Like I can definitely see the DOM being reshaped a little bit in terms of accessibility. Like sometimes you have to write expats that are like so long just to get to a button. Like there should be a better way to do it. And maybe this will drive the change, you know, making it easier for these models to interact with your website. Sharif: There is the Chrome accessibility tree, which is used by screen readers, but a lot of times it's missing a lot of, a lot of useful information. But like in a perfect world, everything would be perfectly annotated for screen readers and we could just use that. That's not the case. [29:30] GPT4 and MultimodalitySwyx: GPT-4 multimodal, has your buddy, Greg, and do you think that that would solve essentially browser agents or desktop agents? Sharif: Greg has not come through yet, unfortunately. But it would make things a lot easier, especially for graphically heavy web pages. So for example, you were using Yelp and like using the map view, it would make a lot of sense to use something like that versus a text based input. Where, how do you serialize a map into text? It's kind of hard to do that. So for more complex web pages, that would make it a lot easier. You get a lot more context to the model. I mean, it seems like that multimodal input is very dense in the sense that it can read text and it can read it really, really well. So you could probably give it like a PDF and it would be able to extract all the text and summarize it. So if it can do that, it could probably do anything on any webpage. Swyx: Yeah. And given that you have some experience integrating Clip with language models, how would you describe how different GPT-4 is compared to that stuff? Sharif: Yeah. Clip is entirely different in the sense that it's really just good at putting images and text into the same latent space. And really the only thing that's useful for is similarity and clustering. Swyx: Like literally the same energy, right? Sharif: Yeah. Swyx: Yeah. And then there's Blip and Blip2. I don't know if you like those. Sharif: Yeah. Blip2 is a lot better. There's actually a new project called, I think, Mini GPT-4. Swyx: Yes. It was just out today. Sharif: Oh, nice. Yeah. It's really cool. It's actually really good. I think that one is based on the Lama model, but yeah, that's, that's like another. Host: It's Blip plus Lama, right? So they, they're like running through Blip and then have Lama ask your, interpret your questions so that you do visual QA. Sharif: Oh, that's cool. That's really clever. Yeah. Ensemble models are really useful. Host: Well, so I was trying to articulate, cause that was, that's, there's two things people are talking about today. You have to like, you know, the moment you wake up, you open Hacker News and go like, all right, what's, what's the new thing today? One is Red Pajama. And then the other one is Mini GPT-4. So I was trying to articulate like, why is this not GPT-4? Like what is missing? And my only conclusion was it just doesn't do OCR yet. But I wonder if there's anything core to this concept of multimodality that you have to train these things together. Like what does one model doing all these things do that is separate from an ensemble of models that you just kind of duct tape together? Sharif: It's a good question. This is pretty related to interoperability. Like how do we understand that? Or how, how do we, why do models trained on different modalities within the same model perform better than two models perform or train separately? I can kind of see why that is the case. Like, it's kind of hard to articulate, but when you have two different models, you get the reasoning abilities of a language model, but also like the text or the vision understanding of something like Clip. Whereas Clip clearly lacks the reasoning abilities, but if you could somehow just put them both in the same model, you get the best of both worlds. There were even cases where I think the vision version of GPT-4 scored higher on some tests than the text only version. So like there might even be some additional learning from images as well. Swyx: Oh yeah. Well, uh, the easy answer for that was there was some chart in the test. That wasn't translated. Oh, when I read that, I was like, Oh yeah. Okay. That makes sense. Sharif: That makes sense. I thought it'd just be like, it sees more of the world. Therefore it has more tokens. Swyx: So my equivalent of this is I think it's a well-known fact that adding code to a language model training corpus increases its ability to do language, not just with code. So, the diversity of datasets that represent some kind of internal logic and code is obviously very internally logically consistent, helps the language model learn some internal structure. Which I think, so, you know, my ultimate test for GPT-4 is to show the image of like, you know, is this a pipe and ask it if it's a pipe or not and see what it does. Sharif: Interesting. That is pretty cool. Yeah. Or just give it a screenshot of your like VS code editor and ask it to fix the bug. Yeah. That'd be pretty wild if it could do that. Swyx: That would be adult AGI. That would be, that would be the grownup form of AGI. [33:30] Sharif's Startup ManualSwyx: On your website, you have this, um, startup manual where you give a bunch of advice. This is fun. One of them was that you should be shipping to production like every two days, every other day. This seems like a great time to do it because things change every other day. But maybe, yeah, tell some of our listeners a little bit more about how you got to some of these heuristics and you obviously build different projects and you iterate it on a lot of things. Yeah. Do you want to reference this? Sharif: Um, sure. Yeah, I'll take a look at it. Swyx: And we'll put this in the show notes, but I just wanted you to have the opportunity to riff on this, this list, because I think it's a very good list. And what, which one of them helped you for Lexica, if there's anything, anything interesting. Sharif: So this list is, it's pretty funny. It's mostly just like me yelling at myself based on all the mistakes I've made in the past and me trying to not make them again. Yeah. Yeah. So I, the first one is like, I think the most important one is like, try when you're building a product, try to build the smallest possible version. And I mean, for Lexica, it was literally a, literally one screen in the react app where a post-process database, and it just showed you like images. And I don't even know if the first version had search. Like I think it did, but I'm not sure. Like, I think it was really just like a grid of images that were randomized, but yeah, don't build the absolute smallest thing that can be considered a useful application and ship it for Lexica. That was, it helps me write better prompts. That's pretty useful. It's not that useful, but it's good enough. Don't fall into the trap of intellectual indulgence with over-engineering. I think that's a pretty important one for myself. And also anyone working on new things, there's often times you fall into the trap of like thinking you need to add more and more things when in reality, like the moment it's useful, you should probably get in the hands of your users and they'll kind of set the roadmap for you. I know this has been said millions of times prior, but just, I think it's really, really important. And I think if I'd spent like two months working on Lexica, adding a bunch of features, it wouldn't have been anywhere as popular as it was if I had just released the really, really boiled down version alongside the stable diffusion release. Yeah. And then there are a few more like product development doesn't start until you launch. Think of your initial product as a means to get your users to talk to you. It's also related to the first point where you really just want people using something as quickly as you can get that to happen. And then a few more are pretty interesting. Create a product people love before you focus on growth. If your users are spontaneously telling other people to use your product, then you've built something people love. Swyx: So this is pretty, it sounds like you've internalized Paul Graham's stuff a lot. Yeah. Because I think he said stuff like that. Sharif: A lot of these are just probably me taking notes from books I found really interesting or like PG essays that were really relevant at the time. And then just trying to not forget them. I should probably read this list again. There's some pretty personalized advice for me here. Oh yeah. One of my favorite ones is, um, don't worry if what you're building doesn't sound like a business. Nobody thought Facebook would be a $500 billion company. It's easy to come up with a business model. Once you've made something people want, you can even make pretty web forms and turn that into a 200 person company. And then if you click the link, it's to LinkedIn for type form, which is now, uh, I think they're like an 800 person company or something like that. So they've grown quite a bit. There you go. Yeah. Pretty web forms are pretty good business, even though it doesn't sound like it. Yeah. It's worth a billion dollars. [38:30] Lexica Aperture V1/2/3Swyx: One way I would like to tie that to the history of Lexica, which we didn't go over, which was just walk us through like Aperture V1, V2, V3, uh, which you just released last week. And how maybe some of those principles helped you in that journey.Sharif: Yeah. So, um, V1 was us trying to create a very photorealistic version of our model of Sable to Fusion. Uh, V1 actually didn't turn out to be that popular. It turns out people loved not generating. Your marketing tweets were popular. They were quite popular. So I think at the time you couldn't get Sable to Fusion to generate like photorealistic images that were consistent with your prompt that well. It was more so like you were sampling from this distribution of images and you could slightly pick where you sampled from using your prompt. This was mostly just because the clip text encoder is not the best text encoder. If you use a real language model, like T5, you get much better results. Like the T5 XXL model is like a hundred times larger than the clip text encoder for Sable to Fusion 1.5. So you could kind of steer it into like the general direction, but for more complex prompts, it just didn't work. So a lot of our users actually complained that they preferred the 1.5, Sable to Fusion 1.5 model over the Aperture model. And it was just because a lot of people were using it to create like parts and like really weird abstract looking pictures that didn't really work well with the photorealistic model trained solely on images. And then for V2, we kind of took that into consideration and then just trained it more on a lot of the art images on Lexica. So we took a lot of images that were on Lexica that were art, used that to train aesthetic models that ranked art really well, and then filtered larger sets to train V2. And then V3 is kind of just like an improved version of that with much more data. I'm really glad we didn't spend too much time on V1. I think we spent about one month working on it, which is a lot of time, but a lot of the things we learned were useful for training future versions. Swyx: How do you version them? Like where do you decide, okay, this is V2, this is V3? Sharif: The versions are kind of weird where you can't really use semantic versions because like if you have a small update, you usually just make that like V2. Versions are kind of used for different base models, I'd say. So if you have each of the versions were a different base model, but we've done like fine tunes of the same version and then just release an update without incrementing the version. But I think when there's like a clear change between running the same prompt on a model and you get a different image, that should probably be a different version. [40:00] Request for AI Startup - LLM ToolsAlessio: So the startup manual was the more you can actually do these things today to make it better. And then you have a whole future page that has tips from, you know, what the series successor is going to be like to like why everyone's genome should be sequenced. There's a lot of cool stuff in there. Why do we need to develop stimulants with shorter half-lives so that we can sleep better. Maybe talk a bit about, you know, when you're a founder, you need to be focused, right? So sometimes there's a lot of things you cannot build. And I feel like this page is a bit of a collection of these. Like, yeah. Are there any of these things that you're like, if I were not building Lexica today, this is like a very interesting thing. Sharif: Oh man. Yeah. There's a ton of things that I want to build. I mean, off the top of my head, the most exciting one would be better tools for language models. And I mean, not tools that help us use language models, but rather tools for the language models themselves. So things like giving them access to browsers, giving them access to things like payments and credit cards, giving them access to like credit cards, giving them things like access to like real world robots. So like, it'd be cool if you could have a Boston dynamic spot powered by a language model reasoning module and you would like to do things for you, like go and pick up your order, stuff like that. Entirely autonomously given like high level commands. That'd be like number one thing if I wasn't working on Lexica. [40:00] Sequencing your GenomeAnd then there's some other interesting things like genomics I find really cool. Like there's some pretty cool things you can do with consumer genomics. So you can export your genome from 23andMe as a text file, like literally a text file of your entire genome. And there is another tool called Prometheus, I think, where you upload your 23andMe text file genome and then they kind of map specific SNPs that you have in your genome to studies that have been done on those SNPs. And it tells you really, really useful things about yourself. Like, for example, I have the SNP for this thing called delayed sleep phase disorder, which makes me go to sleep about three hours later than the general population. So like I used to always be a night owl and I never knew why. But after using Prometheus it pretty much tells you, oh, you have the specific genome for specific SNP for DSPS. It's like a really tiny percentage of the population. And it's like something you should probably know about. And there's a bunch of other things. It tells you your likelihood for getting certain diseases, for certain cancers, oftentimes, like even weird personality traits. There's one for like, I have one of the SNPs for increased risk taking and optimism, which is pretty weird. That's an actual thing. Like, I don't know how. This is the founder gene. You should sequence everybody. It's pretty cool. And it's like, it's like $10 for Prometheus and like 70 bucks for 23andMe. And it explains to you how your body works and like the things that are different from you or different from the general population. Wow. Highly recommend everyone do it. Like if you're, if you're concerned about privacy, just purchase a 23andMe kit with a fake name. You don't have to use your real name. I didn't use my real name. Swyx: It's just my genes. Worst you can do is clone me. It ties in with what you were talking about with, you know, we want the future to be like this. And like people are building uninspired B2B SaaS apps and you and I had an exchange about this. [42:00] Believe in Doing Great ThingsHow can we get more people to believe they can do great things? Sharif: That's a good question. And I like a lot of the things I've been working on with GP3. It has been like trying to solve this by getting people to think about more interesting ideas. I don't really know. I think one is just like the low effort version of this is just putting out really compelling demos and getting people inspired. And then the higher effort version is like actually building the products yourself and getting people to like realize this is even possible in the first place. Like I think the baby AGI project and like the GPT Asian projects on GitHub are like in practice today, they're not super useful, but I think they're doing an excellent job of getting people incredibly inspired for what can be possible with language models as agents. And also the Stanford paper where they had like the mini version of Sims. Yeah. That one was incredible. That was awesome. Swyx: It was adorable. Did you see the part where they invented day drinking? Sharif: Oh, they did? Swyx: Yeah. You're not supposed to go to these bars in the afternoon, but they were like, we're going to go anyway. Nice. Sharif: That's awesome. Yeah. I think we need more stuff like that. That one paper is probably going to inspire a whole bunch of teams to work on stuff similar to that. Swyx: And that's great. I can't wait for NPCs to actually be something that you talk to in a game and, you know, have their own lives and you can check in and, you know, they would have their own personalities as well. Sharif: Yeah. I was so kind of off topic. But I was playing the last of us part two and the NPCs in that game are really, really good. Where if you like, point a gun at them and they'll beg for their life and like, please, I have a family. And like when you kill people in the game, they're like, oh my God, you shot Alice. Like they're just NPCs, but they refer to each other by their names and like they plead for their lives. And this is just using regular conditional rules on NPC behavior. Imagine how much better it'd be if it was like a small GPT-4 agent running in every NPC and they had the agency to make decisions and plead for their lives. And I don't know, you feel way more guilty playing that game. Alessio: I'm scared it's going to be too good. I played a lot of hours of Fallout. So I feel like if the NPCs were a lot better, you would spend a lot more time playing the game. Yeah. [44:30] Lightning RoundLet's jump into lightning round. First question is your favorite AI product. Sharif: Favorite AI product. The one I use the most is probably ChatGPT. The one I'm most excited about is, it's actually a company in AI grants. They're working on a version of VS code. That's like an entirely AI powered cursor, yeah. Cursor where you would like to give it a prompt and like to iterate on your code, not by writing code, but rather by just describing the changes you want to make. And it's tightly integrated into the editor itself. So it's not just another plugin. Swyx: Would you, as a founder of a low code prompting-to-code company that pivoted, would you advise them to explore some things or stay away from some things? Like what's your learning there that you would give to them?Sharif: I would focus on one specific type of code. So if I'm building a local tool, I would try to not focus too much on appealing developers. Whereas if I was building an alternative to VS code, I would focus solely on developers. So in that, I think they're doing a pretty good job focusing on developers. Swyx: Are you using Cursor right now? Sharif: I've used it a bit. I haven't converted fully, but I really want to. Okay. It's getting better really, really fast. Yeah. Um, I can see myself switching over sometime this year if they continue improving it. Swyx: Hot tip for, for ChatGPT, people always say, you know, they love ChatGPT. Biggest upgrade to my life right now is the, I forked a menu bar app I found on GitHub and now I just have it running in a menu bar app and I just do command shift G and it pops it up as a single use thing. And there's no latency because it just always is live. And I just type, type in the thing I want and then it just goes away after I'm done. Sharif: Wow. That's cool. Big upgrade. I'm going to install that. That's cool. Alessio: Second question. What is something you thought would take much longer, but it's already here? Like what, what's your acceleration update? Sharif: Ooh, um, it would take much longer, but it's already here. This is your question. Yeah, I know. I wasn't prepared. Um, so I think it would probably be kind of, I would say text to video. Swyx: Yeah. What's going on with that? Sharif: I think within this year, uh, by the end of this year, we'll have like the jump between like the original DALL-E one to like something like mid journey. Like we're going to see that leap in text to video within the span of this year. Um, it's not already here yet. So I guess the thing that surprised me the most was probably the multi-modality of GPT four in the fact that it can technically see things, which is pretty insane. Swyx: Yeah. Is text to video something that Aperture would be interested in? Sharif: Uh, it's something we're thinking about, but it's still pretty early. Swyx: There was one project with a hand, um, animation with human poses. It was also coming out of Facebook. I thought that was a very nice way to accomplish text to video while having a high degree of control. I forget the name of that project. It was like, I think it was like drawing anything. Swyx: Yeah. It sounds familiar. Well, you already answered a year from now. What will people be most surprised by? Um, and maybe the, uh, the usual requests for startup, you know, what's one thing you will pay for if someone built it? Sharif: One thing I would pay for if someone built it. Um, so many things, honestly, I would probably really like, um, like I really want people to build more, uh, tools for language models, like useful tools, give them access to Chrome. And I want to be able to give it a task. And then just, it goes off and spins up a hundred agents that perform that task. And like, sure. Like 80 of them might fail, but like 20 of them might kind of succeed. That's all you really need. And they're agents. You can spin up thousands of them. It doesn't really matter. Like a lot of large numbers are on your side. So that'd be, I would pay a lot of money for that. Even if it was capable of only doing really basic tasks, like signing up for a SAS tool and booking a call or something. If you could do even more things where it could have handled the email, uh, thread and like get the person on the other end to like do something where like, I don't even have to like book the demo. They just give me access to it. That'd be great. Yeah. More, more. Like really weird language model tools would be really fun.Swyx: Like our chat, GPT plugins, a step in the right direction, or are you envisioning something else? Sharif: I think GPT, chat GPT plugins are great, but they seem to only have right-only access right now. I also want them to have, I want these like theoretical agents to have right access to the world too. So they should be able to perform actions on web browsers, have their own email inbox, and have their own credit card with their own balance. Like take it, send emails to people that might be useful in achieving their goal. Ask them for help. Be able to like sign up and register for accounts on tools and services and be able to like to use graphical user interfaces really, really well. And also like to phone home if they need help. Swyx: You just had virtual employees. You want to give them a Brex card, right? Sharif: I wouldn't be surprised if, a year from now there was Brex GPT or it's like Brex cards for your GPT agents. Swyx: I mean, okay. I'm excited by this. Yeah. Kind of want to build it. Sharif: You should. Yeah. Alessio: Well, just to wrap up, we always have like one big takeaway for people, like, you know, to display on a signboard for everyone to see what is the big message to everybody. Sharif: Yeah. I think the big message to everybody is you might think that a lot of the time the ideas you have have already been done by someone. And that may be the case, but a lot of the time the ideas you have are actually pretty unique and no one's ever tried them before. So if you have weird and interesting ideas, you should actually go out and just do them and make the thing and then share that with the world. Cause I feel like we need more people building weird ideas and less people building like better GPT search for your documentation. Host: There are like 10 of those in the recent OST patch. Well, thank you so much. You've been hugely inspiring and excited to see where Lexica goes next. Sharif: Appreciate it. Thanks for having me. Get full access to Latent Space at www.latent.space/subscribe

Commentary by Dr. Valentin Fuster

During this episode of SurgOnc Today®, Valerie P. Grignol, MD, Vice Chair of the SSO Sarcoma Disease Site Work Group, Timur Mitin, MD, PhD, Arun Singh, MD, and Neha Goel, MD, discuss the evaluation, treatment and management of primary and radiation induced angiosarcoma of the breast through a case based discussion.

Commentary by Dr. Valentin Fuster

In this sixth episode of Conjuncture, Jordan T. Camp's guest is Leniqueca A. Welcome. They discuss her ethnographic research in Trinidad, what she calls the “tyranny of the present,” and the possibilities of abolitionist politics today. Conjuncture is a monthly web series and podcast curated and co-produced by Jordan T. Camp and Christina Heatherton for the Trinity Social Justice Initiative. It features interviews with activists, artists, scholars, and public intellectuals. Taking its title from Antonio Gramsci and Stuart Hall's conceptualization, it highlights intellectual work engaged in struggles over the meaning and memory of particular historical moments. Amidst a global crisis of hegemony, this web series curates conversations about the burning questions of the conjuncture. Leniqueca A. Welcome is an assistant professor of International and Urban Studies at Trinity College. Her work appears in influential venues such as Small Axe, American Anthropologist, Multimodality and Society, Inquiry, City and Society, as well as edited volumes including Sovereignty Unhinged (Duke, forthcoming). She is currently working on her first book, Come Out of This World: Beyond Terrains of Criminalization to Where Life is Precious. Jordan T. Camp is an Assistant Professor of American Studies and Co-Director of the Social Justice Initiative at Trinity College.

CardioNerds (Daniel Ambinder), ACHD series co-chairs,  Dr. Josh Saef (ACHD fellow, University of Pennsylvania) Dr. Daniel Clark (ACHD fellow, Vanderbilt University), and ACHD FIT lead Dr. Jon Kochav (Columbia University) join Dr. Eric Krieger (Director of the Seattle Adult Congenital Heart Service and the ACHD Fellowship, University of Washington) to discuss multimodality imaging in congenital heart disease. Audio editing by CardioNerds Academy Intern, Dr. Maryam Barkhordarian. In this episode we discuss the strengths and weaknesses of the imaging modalities most commonly utilized in the diagnosis and surveillance of patients with ACHD.  Specifically, we discuss transthoracic and transesophageal echocardiography, cardiac MRI and cardiac CT. The principles learned are then applied to the evaluation of two patient cases – a patient status post tetralogy of Fallot repair with a transannular patch, and a patient presenting with right ventricular enlargement of undetermined etiology. The CardioNerds Adult Congenital Heart Disease (ACHD) series provides a comprehensive curriculum to dive deep into the labyrinthine world of congenital heart disease with the aim of empowering every CardioNerd to help improve the lives of people living with congenital heart disease. This series is multi-institutional collaborative project made possible by contributions of stellar fellow leads and expert faculty from several programs, led by series co-chairs, Dr. Josh Saef, Dr. Agnes Koczo, and Dr. Dan Clark. The CardioNerds Adult Congenital Heart Disease Series is developed in collaboration with the Adult Congenital Heart Association, The CHiP Network, and Heart University. See more Disclosures: None Pearls • Notes • References • Guest Profiles • Production Team CardioNerds Adult Congenital Heart Disease PageCardioNerds Episode PageCardioNerds AcademyCardionerds Healy Honor Roll CardioNerds Journal ClubSubscribe to The Heartbeat Newsletter!Check out CardioNerds SWAG!Become a CardioNerds Patron! Pearls - Cardiovascular Multimodality Imaging in Congenital Heart Disease Transthoracic echocardiography (TTE) is the first line diagnostic test for the diagnosis and surveillance of congenital heart disease due to widespread availability, near absent contraindications, and ability to perform near comprehensive structural, functional, and hemodynamic assessments in patients for whom imaging windows allow visualization of anatomic areas of interest.Transesophageal echocardiography (TEE) use in ACHD patients is primarily focused on similar indications as in acquired cardiovascular disease patients: the assessment of endocarditis, valvular regurgitation/stenosis severity and mechanism, assessment of interatrial communications in the context of stroke, evaluation for left atrial appendage thrombus, and for intraprocedural guidance. When CT or MRI are unavailable or contraindicated, TEE can also be used when transthoracic imaging windows are poor, or when posterior structures (e.g. sinus venosus, atrial baffle) need to be better evaluated.Cardiac MRI (CMR) with MR angiography imaging is unencumbered by imaging planes or body habitus and can provide comprehensive high resolution structural and functional imaging of most cardiac and extracardiac structures. Additional key advantages over echocardiography are ability to reproducibly quantify chamber volumes, flow through a region of interest (helpful for quantifying regurgitation or shunt fraction), assess for focal fibrosis via late gadolinium enhancement imaging, and assess the right heart.Cardiac CT has superior spatial resolution in a 3D field of view which makes it useful for clarifying anatomic relationships between structures, visualizing small vessels such as coronary arteries or collateral vessels, and assessing patency of larger vessels (e.g branch pulmonary arteries) through metallic stents which may obscure MR imaging.