Podcasts about Baseline

Send us Fan MailOn est back! Retour d'IA Café. Saison 7. Au programme : rattrapage de l'actualité IA des deux derniers mois…en attendant l'IApocalypse!! Au programme L'apocalypse annoncé de Jacob Coxon et AnthropicDémocratie, état de droit et pouvoir économique : Qui doit définir les règles de l'IA? Claude (Anthropic) promet de filigraner le texte généré par l'IAAttaque orchestrée par des agents IA autonomes d'Open AI contre Hugging face. OpenAI annonce une pause sur ses modèles frontières. Et (trois semaines plus tard) lance son nouveau modèle ChatGPT 6-Astra! Êtes-vous capable de détecter les textes IA? Vraiment?  « Bot or not: Can people tell the difference between stories written by a human or by an AI system? Une IA pour créer des virus - Des scientifiques de l'Université Stanford qui ont utilisé des IA pour créer de vrais nouveaux types de virus. SOCAN vs Suno - Poursuite judiciaire : (Société canadienne des auteurs-compositeurs et éditeurs de musique) vs Suno!  Claude Fable 5 -Mythos; Grok 4.5; Kimi k3, DeepSeek V4ChatGPT work en réponse à Claude coworkBonne écoute.Production et animation: Jean-François Sénéchal, Ph.D.BaristIAs:  Sylvain Munger Ph.D, Marie-Ève Vachon Savary, Catherine Légaré-Pelletier et Jean-François Sénéchal.  Collaborateurs et collaboratrices:  Véronique Tremblay, Stéphane Mineo, Frédérick Plamondon Ph.D., Shirley Plumerand, Sylvain Munger Ph.D, Marie-Ève Vachon Savary, Louis Cormier, Catherine Légaré-Pelletier. En musique : PP2, Aubert Sénéchal (2025) (c)BaselineVoyez ce que Baseline peut faire pour vous !Institut Intelligence et Données (IID) Un pôle d'expertise au cœur de l'écosystème IA québécoisDisclaimer: This post contains affiliate links. If you make a purchase, I may receive a commission at no extra cost to you.Support the show

Derek Evans managed a $6 billion REIT lending portfolio at Wells Fargo and structured a $3.5 billion bridge loan before joining Realberry as CFO. Now he’s helping guide the Realberry investment strategy through a move from $300 million to over $1 billion in annual deals. Chris Lopez sits down with Derek in studio to unpack how one of Colorado’s most established real estate firms actually decides what to build, what to buy, and what to walk away from. Derek spent 22 years at Wells Fargo before Chad McWhinney recruited him during COVID. His job now is to institutionalize the firm behind Union Station, Dairy Block, and Centerra without losing what made it work in the first place. The Realberry investment strategy is built on conservative leverage and a layered capital stack. Each investor tier fits a different kind of deal, and the firm is opening up a new digital channel to reach accredited investors for the first time. Derek walks through how those pieces fit together. From there, the conversation turns to the investment committee that decides what moves forward. Eight members, one meeting a week, and a voting rule that has ended more Denver metro deals than most investors would guess. Water rights, business climate, and market fundamentals across Colorado, Austin, and Phoenix all shape what gets a green light. Derek closes with where he sees capital moving next: luxury hospitality driven by K-shaped wealth trends, mixed-use in Loveland and Baseline, the firm’s first 55+ for-rent community, and a downtown Denver thesis that still hinges on getting employees back in office seats. In This Episode We Cover: Why Realberry targets 60 to 65% leverage and puts in 5 to 10% GP capital How ultra-high net worth, family office, and institutional capital each fit different deals The two no vote rule that kills any deal at investment committee Why the firm is actively bidding on multifamily in Colorado, Austin, and Phoenix What Derek looks for in luxury hospitality and 55+ for-rent product Why water rights have killed multiple Denver metro deals in the last five years The Realberry investment strategy behind downtown Denver, office-to-resi, and hotel conversions This episode wraps our three-part Realberry series. If you missed the earlier conversations with Chad McWhinney and Taylor Hazlett, go back and start there for the full picture of how the firm thinks about capital, deals, and Colorado. Watch the Youtube Video https://youtu.be/DvQ4uDlSv5Y Timestamps 00:00 Welcome and guest introduction 01:41 Managing a $6 billion REIT lending portfolio at Wells Fargo 04:49 Unsecured vs secured real estate lending (at a larger level*)  06:17 Why Derek left Wells Fargo for Realberry during COVID 08:06 Growing Realberry from $300M to $1B in annual deals 11:55 The Realberry capital stack and conservative leverage 14:45 Ultra-high net worth, family office, and accredited investor tiers 15:55 Institutional capital and the control rights tradeoff 22:19 Inside the Realberry investment committee 37:30 How two no votes kill any deal 28:47 Why Realberry is bidding on multifamily right now 31:54 Colorado, Austin, Phoenix, and Nashville fundamentals 32:25 Legislation, affordability, and Colorado’s business climate 37:11 Placing capital in Colorado 40:47 Water rights and the deals Realberry has killed 44:12 Luxury hospitality and the K-shaped wealth trend 45:29 Downtown Denver and office-to-resi conversions Links in Podcast Realberry Website: https://www.realberry.com Portfolio: https://www.realberry.com/portfolio LinkedIn: https://www.linkedin.com/company/realberryinvest Instagram: https://www.instagram.com/realberryinvest Investor inquiries: ir@realberry.com Derek Evans LinkedIn: https://www.linkedin.com/in/derek-evans-051b80a5/ This is the third and final episode in our three-part series with Realberry. If you haven’t caught the first two, start here. Episode 1 — Chad McWhinney, Co-Founder and CEO. Chad walks through the origin story, from a berry stand in Loveland to Union Station and Centerra. He covers 35 years of building in Colorado, how the firm thinks about long-hold capital, and the vision behind the rebrand to Realberry. Episode 2 — Taylor Hazlett, Senior Director of Private Capital. Taylor breaks down how Realberry filters 100 deals to close 2 or 3. He covers the firm’s multifamily underwriting in today’s market, the gap between replacement cost and acquisition pricing, and early signs of recovery across the Front Range. Together, the three episodes give you the founder’s vision, the deal team’s discipline, and the CFO’s capital architecture. That’s the full picture of how Realberry works. Who is Realberry? Realberry, formerly McWhinney, is a Denver-based real estate investment, development, and management firm founded in 1991 by brothers Chad and Troy McWhinney. For nearly 35 years, the firm has focused on creating places people love, with a portfolio spanning master-planned communities, multifamily, hospitality, industrial, and mixed-use developments. Its work includes Denver Union Station, Dairy Block, the Crawford Hotel, and Centerra, and has earned ULI Awards of Excellence, Michelin Keys, and U.S. News Best Hotels recognition. Realberry is family-founded, community-centered, and future-focused.

Commentary by Dr. Jian'an Wang.

Budgeting for irregular income can feel difficult when your creative work changes from month to month. One month you may be fully booked, selling commissions, performing in busy venues or finishing a strong run of work. The next month may feel quiet. For artists, freelancers and creative business owners, that income rollercoaster can create stress, uncertainty and financial anxiety. In this episode, we share three simple steps to help you build a budget that fits a creative lifestyle: work out your baseline expenses, build a buffer for slower months, and use a simple flexible budgeting system. About this episode Irregular income is normal for many creatives. Work can arrive in waves. Projects may come in quickly, then slow down. Performances, commissions, client work, funding and seasonal demand can all affect what comes into your bank account. That does not mean budgeting is impossible. It means your budget needs to reflect the reality of your income pattern. In this episode, we focus on a simple approach that helps you understand what you need each month, prepare for quieter periods and keep enough flexibility to enjoy life while staying financially responsible. Why this matters When income is unpredictable, it is easy to feel out of control. You may spend more in a good month, only to feel under pressure when work slows down. A budget gives you a clearer plan. It helps you see what must be covered, what can wait, and what should be put aside for later. Budgeting is not about removing freedom from your creative life. Used well, it gives you more freedom because you know where you stand. “With a little bit of planning, you can make it work for you instead of against you.” Key points from this episode Start with your baseline expenses Your baseline expenses are the essentials you need to cover every month, whether work is busy or quiet. These include things such as rent, food, utilities, internet and the other basics that keep you going. Write them down clearly. Use a notebook, a notes app, a spreadsheet, your bank statements or your credit card statements. The point is to find your monthly survival number. In the episode, we use an example of £1,500 per month. That figure becomes the target you need to cover before anything else. Knowing your baseline gives you a solid foundation. Instead of guessing, you know the minimum amount you need to keep your head above water. Build a buffer for slower months Once you know your baseline, the next step is to build a buffer. A buffer is a financial cushion. It helps you cover quieter months when income slows down. Every business has busier periods and quieter periods, and creative businesses are no different. During stronger months, get into the habit of putting something aside. It might be £50, £100, or another amount that works for you. Small regular amounts build up over time. A useful target is to work towards three months of baseline expenses. If your baseline is £1,500 per month, the target buffer would be £4,500. That may sound difficult, but it does not need to happen overnight. Consistent small steps matter. This links closely with broader cash planning. Our episode on Cash Flow Management Tips gives wider support for building resilience and staying prepared. Use a simple flexible budgeting system A budget should not feel like a straitjacket. It should be a discipline that helps you make better choices. One simple approach is to divide your income into three categories: Essentials:rent, bills, food and the non-negotiables.Fun:things that support your life, energy and creativity.Savings:your buffer, long-term goals, training, equipment, projects or time out. The strength of this system is flexibility. In one month, you may put more towards fun because work has gone well. In another month, you may focus on rebuilding your buffer. The goal is not perfection. The goal is awareness, consistency and control. Why budgeting helps creative confidence When your income is unpredictable, your numbers can feel emotional. A quiet month may feel like failure. A busy month may create a false sense of security. Budgeting helps you see the bigger picture. It separates short-term emotion from practical planning. Once you know your baseline, track your income and understand your buffer, you can make calmer decisions. You can plan your spending, protect your essentials and avoid being surprised by every quiet period. Our episode on Bookkeeping for Small Business is a helpful next step if you want to build the habit of tracking what comes in and what goes out. Your two simple actions There are two practical actions to take from this episode. 1. Work out your baseline Write down your essential monthly expenses. Be honest. Use actual bank and card information where possible, rather than guessing. 2. Start tracking Track your income and expenses regularly. It does not need to be fancy. The important thing is to start connecting with your numbers. Once you know what is coming in and what is going out, your confidence grows and your anxiety can reduce. FAQs How do you budget with irregular income? Start by working out your baseline expenses. Then build a buffer for slower months and use a flexible budgeting system that separates essentials, fun and savings. What are baseline expenses? Baseline expenses are the essential costs you need to cover each month. They usually include rent, food, utilities, internet and other non-negotiable living or business costs. How much should creatives keep as a buffer? A useful target is three months of baseline expenses. This is a practical planning guide, not a fixed rule. Start with small regular amounts and build the buffer over time. Does budgeting restrict creativity? No. A good budget should support creativity, not restrict it. It helps you understand what you can afford, prepare for quiet months and make decisions with less stress. What should creatives track each month? Track what income comes in, what expenses go out, what essentials must be covered, and what amount can be saved towards your buffer or longer-term goals. Episode Timecodes 00:00 – Budgeting when creative income feels like a rollercoaster00:18 – Why irregular income is common for creatives00:51 – The three-step budgeting game plan01:00 – Working out your baseline expenses01:46 – Using bank statements to find your monthly target02:05 – Building a buffer for quieter months02:45 – Saving small amounts during busier months03:24 – Why consistency matters03:43 – Using a simple flexible budget04:00 – Essentials, fun and savings04:59 – Two practical actions to take next05:38 – Budgetwhizz and planning support Related episodes Cash Flow Management Tips for Small BusinessesBuild Your Cash Flow with a SpreadsheetBookkeeping for Small Business: Your Numbers Tell a Story Key takeaway Budgeting for irregular income does not need to be a headache. Start with your baseline expenses, build a buffer for quieter months, and use a simple flexible system that allows for essentials, fun and savings. With the right habits and the right tools, you can feel more in control and give yourself more freedom to focus on the creative work you love. About the Podcast The I Hate Numbers podcast, presented by Mahmood Reza, helps business owners understand accounting, tax, finance, profit, cash flow and business planning in a practical way. We simplify financial topics so you can make better decisions and feel more confident with your numbers. You can also watch more practical finance and tax support on the

Josh is away on his honeymoon, so my brother Jared joins me for this week's episode of the podcast. We recap week one, but we had to focus on the absolute robbery of the Western Michigan/Michigan game. We detail how we both agree that Michigan lost that game, but we could be a second too late on that opinion. Then Jared gives his power rankings and heisman watch list.

Shifting Baseline Syndrome ist ein hübsch klingender Ausdruck für einen perfiden Mechanismus: Jede Generation hält den Zustand der Welt, den sie vorfindet, für den Normalzustand. Egal, wie sehr diese Welt bereits ausgedünnt, ausgeräumt oder heruntergewirtschaftet ist. Und auch jede Generation passt ihre Ohren an die dominante Klangumgebung an und kaum jemand merkt, dass sich die Baseline verschiebt. Erst wenn Du eine Aufnahme hörst, die 50 Jahre alt ist, wirkt sie plötzlich wie eine Stimme aus einem anderen Planeten.

It's a Modest Proposal Tuesday with Judd Zulgad joining Thor Nystrom to discuss proposals for Kyler Murray to be at a baseline level of play in order for the team to have success and for fans to set their expectations and stick to them for end of season consequences. Plus a talk about if Harrison Smith will suit up on Sunday!See Privacy Policy at https://art19.com/privacy and California Privacy Notice at https://art19.com/privacy#do-not-sell-my-info.

Keith breaks down the "baseline trap" in investor psychology, showing how rising income and lifestyle creep can quietly undermine the feeling of financial freedom.  He then shares a grounded outlook for U.S. home prices, outlining how inflation, AI-driven job growth, limited inventory, and strong homeowner equity are shaping the market.  He closes with a data-driven look at where population growth is heading through 2040, especially in Texas and Florida, and what that could mean for long-term real estate demand and investing strategy. Episode Page: GetRichEducation.com/622 For access to properties or free help with a GRE Investment Coach, start here: GREmarketplace.com GRE Free Investment Coaching: GREinvestmentcoach.com Get mortgage loans for investment property: RidgeLendingGroup.com or call 855-74-RIDGE  or e-mail: info@RidgeLendingGroup.com Invest with Freedom Family Investments.  For predictable 10-12% quarterly returns, visit FreedomFamilyInvestments.com/GRE or text  FAMILY to 66866  Join Mid South Home Buyers' one-time, free live webinar featuring Keith Weinhold on September 30 at GetRichEducation.com/MidSouth to learn how Memphis' economic expansion could create new real estate investment opportunities, and have your questions answered in real time. Will you please leave a review for the show? I'd be grateful. Search "how to leave an Apple Podcasts review"  For advertising inquiries, visit: GetRichEducation.com/ad Best Financial Education: GetRichEducation.com Get our wealth-building newsletter free— GREletter.com  Our YouTube Channel: www.youtube.com/c/GetRichEducation Follow us on Instagram: @getricheducation Complete episode transcript:   Keith Weinhold  0:01   Welcome to GRE. I'm your host Keith Weinhold. Investor psychology often falls into the baseline trap. Learn what's going to happen to home prices over the next year. Then more than half of America's population growth until 2040 will occur in just these two states. All today on Get Rich Education. What if I told you that one of America's strongest cash flow real estate markets is also becoming the new brains and brawn behind AI? That city is Memphis, believe it or not. In September 30th, we're going to show you why the smart money is paying attention now, along with an investing opportunity you won't want to miss. Join me, Terry Kerr and Matthew Van Horn of Mid South Homebuyers, the largest turnkey company in Memphis with more than 6000 homes under management, for a free live webinar the likes of which I've never done before. We're going to look at what billions in new investment could mean for jobs, housing demand, neighborhood appreciation, and your portfolio. Everyone who attends live will also get exclusive access to the best deal terms Mid South has ever offered. Reserve your free seat at getricheducation.com/midsouth again. that september 30. Don't say we didn't tell you. Save your spot at getricheducation.com/midsouth   Speaker 1  1:34   You're listening to the show that has created more financial freedom than nearly any show in the world. This is Get Rich Education.   Keith Weinhold  1:50   Welcome to GRE from Jackson Hole, Wyoming, to Jackson, Mississippi, and across 188 nations worldwide. I'm Keith Weinhold. This is Get Rich Education, and Happy Labor Day. Let's talk about your investor psychology, because as you grow your wealth and your portfolio size, there is a trap that you will almost certainly fall into, and I'm not infallible. I've fallen into this trap to some extent too. That is the baseline trap. It's the tendency for every improvement in your income, your wealth, or your lifestyle to become your new normal. Once this happens, the improvement stops feeling like progress, and you need even more just to feel equally successful, if you get used to flying first class and then you have to drop back to coach again, it feels less like flying and more like being deported. Psychologically, we fall into the baseline trap because the human mind evaluates Life relatively, not absolutely. We don't simply ask ourselves how good is my life, how good is my situation. Instead, we ask how does this compare with what I've recently experienced, what I expected, and what others have, and there are a number of forces that drive the baseline trap. One is hedonic adaptation. Hedonic means pleasure seeking. People rapidly adjust to improvements. The first month of receiving a new $5,000 in passive income that feels transformative. After two years, it feels completely ordinary. The income didn't become less valuable. Your nervous system simply stopped registering it as new. Yesterday's luxury became today's wallpaper. A force driving the baseline trap is a shifting reference point. Gains and losses are measured against a mental baseline. Once your portfolio reaches, say, a $2 million net worth, well, your mind soon begins treating the $2 million as mine. You're like, hey, this is mine now, even if much of it came from recent appreciation. A decline to 1.8 million, therefore, feels like losing 200k rather than still having substantially more wealth than you did just a few years ago. Well, instead, you're only focused on the 200k paper loss. Then there's loss aversion psychologically. Losses generally hurt more than equivalent gains feel good. After a higher standard becomes normal, surrendering and. Any part of it feels like some blood-curdling loss. That's why reducing spending from 20k to 15k per month that can feel painful, even if 15k once felt luxurious to you.   Keith Weinhold  5:16   There's also the lifestyle creep component. People convert variable gains into fixed commitments. What do I mean? I mean like a strong income year. Oh, pretty soon that becomes a larger mortgage. Rental cash flow that becomes a vehicle payment. A bonus that becomes private school tuition, portfolio appreciation. Well, that supports new borrowing. See, pleasures that were once optional have now become obligations. And you got to ask, wait, how did that happen to you? You're supposed to have a life of options and not obligations. That's what financial freedom is supposed to be. The baseline then is no longer merely psychological; it becomes embedded in real monthly expenses. Then there's also the dangerous driver of the baseline trap that's called, oh no, social comparison. We commonly measure success against our peers, but instead, what you should do is measure it against your former self. Because as you become wealthier, see your comparison group changes too. If you've got five rentals, you soon stop comparing yourself with someone that owns none, you might even begin comparing yourself with people who own 50 of them, and why not? It's natural, after all. That is where you want to go, despite enormous progress. See, that's how you can feel left further behind. Then there's the recency bias. Your mind gives enormously disproportionate weight to recent experience. A few years of 15% returns, like what happened in 2021 and 2022 in real estate. Oh, you could begin expecting 15% after rapidly appreciating real estate, continued appreciation feels normal. A favorable cycle gets mistaken for the natural baseline, and then when conditions normalize, ordinary performance feels rather defective. Then there's identity inflation. That's a trap. This is when accomplishments become woven into your very identity, like I'm a multi-million-dollar entrepreneur, or I own 20 properties, or my income always grows. Okay, once success becomes identity, maintaining the baseline feels necessary just to preserve your self worth. Now, with this condition, see a temporary setback. It doesn't merely affect the numbers.   Keith Weinhold  8:08   It feels like evidence that you're becoming a lesser person, and the brain rewards progress more than possession. Humans are energized by movement toward a goal, reaching the goal often produces less lasting satisfaction than you expect. Buying the 10th rental creates a dopamine hit, and owning it three years later does not. The investor therefore creates another target, not always because another property is even needed, but because continued pursuit restores the feeling of progress, success erases the memory of constraint. As your wealth grows, it becomes difficult to remember emotionally what financial insecurity even felt like I mean you might intellectually remember earning 60k, but you no longer experience today's 300k income in comparison with it. Your comparison point quietly changes from your former life to your best recent year. The paradox is that your circumstances improve faster than your experience of them? The goal is not to stop growing; it is to prevent every improvement from becoming a new psychological necessity. Keep growing your means, but don't let success redefine enough every time you achieve it, don't let it redefine enough. Let's say you acquire rentals and you do generate another 5k per month. The trap is that your spending and expectations gradually rise by 5k. You're wealthier, but you don't. Don't feel freer. Instead of investments buying freedom, they merely finance a more expensive baseline, and it can distort how you view your portfolio. 10 properties once felt like an extraordinary accomplishment, and soon 10 feels ordinary, and 20 becomes necessary. You keep moving the finish line, and this is closely related to hedonic adaptation and lifestyle creep. But it extends beyond spending because your definition of enough keeps on rising. So the antidote certainly is not living small forever-it's deliberately separating the growth rates of your assets and your lifestyle. What you want to do is grow your means faster than you grow your baseline. Really, that's the key. You're gonna be more satisfied. Instead of simply living below your means, you sure do want to grow your means, but don't let every gain become a permanent new obligation. Let some additional cash flow purchase you things like time, resilience, and optionality-not merely nicer recurring expenses. If your lifestyle rises as fast as your passive income, you're wealthier, but no freer.   Keith Weinhold  11:28   So here's what you do: when your income rises, let your lifestyle rise about half that much. Otherwise, if you upgrade your lifestyle too much, say that you receive an extra $3,000 in monthly rental income, then you add in a luxury car payment, better vacations, and more expensive restaurants. Pretty soon, that extra 3k that feels necessary instead of liberating, and then there's also the record income comparison part of the trap. Say your business earns $1 million during an exceptional year. The next year, it earns a still impressive 850k, but you experience it as failure because the unusually strong year became your new baseline. Don't let that happen. You can compare yourself to others that can be motivating, but the more important comparison is to the former you. Now, another way that investors fall into the baseline trap in real estate is how an exceptional market becomes the standard. Say that you bought rental properties in 2012. Well, 2012 was perhaps the best time to buy real estate in generations. This was shortly after the global financial crisis, so there was this confluence of low prices, low interest rates, strong cash flow, and you had little competition as well. I mean, you had it all in 2012, and those deals performed spectacularly in today's market. Available properties produce lower initial cash flow, but they could still deliver respectable total returns through appreciation, rent income, principal paydown, tax benefits, and inflation profiting. But a losing investor rejects all of those things because they aren't as attractive as the once-in-a-generation deals of 2012, or even the rock-bottom low-rate days of 2020, they fell into the baseline trap. The trap here is that an unusually favorable period for real estate became the new benchmark. It's sort of like how last week I told you about how the deal structure always changes over time from the Reagan administration until today. Today the deal is with Burr properties, and it's also with buying new builds with rate buydowns. But see, in 2012 there were almost zero available new build properties that were created for investors to rent to others.   Keith Weinhold  14:25   Over time, with these new builds that you're adding now, you're going to have fewer maintenance and repair expenses. Tenants tend to stay in new builds longer, and new builds appreciate better over the long run. See, I wasn't getting any of those benefits in 2012, and I bought rental real estate in 2012, and I bought real estate recently as well. Not falling into the baseline trap, because today it's still difficult to find any investment bet. Than residential real estate with a loan, it is a scarce asset that people are going to continue to need. So here we are today, about 15 years on from 2012. Water market conditions like now. Let's talk about that and what can we expect for the next year? National home prices keep rising, but they're only about one half of 1% higher than they were a year ago. I mean, that's an appreciation level with the enthusiasm of someone attending a seven a.m. meeting. I do expect national home prices to keep rising modestly over the next year. Let me tell you about why, and then what the drivers are. And to be clear, we're talking about single-family homes up to fourplexes here. I'll discuss apartments later today. Well, the drivers for continued price growth are many of the same reasons that home prices are up just a little since last year. There are four of them. These four are inflation, the AI boom, short inventory, and a lack of distressed sellers. So let's unpack all of these four factors that I've identified for putting a floor underneath home prices, inflationary pressure is poised to raise replacement cost, energy, wages, and tariffs make those inputs more expensive, and the more war we have, the more inflation we have. A home is a bundle of land, labor, lumber, concrete, copper, and all sorts of energy inputs, plus 14 trips to Home Depot because someone forgot the correct nails and screws. That's what a home is. Recent home price growth it has lagged today's 3.4% CPI inflation rate. So again, we're not even talking about inflation-adjusted gains here. AI that creates local housing heat. It's not so much a nationwide driver of home prices. And in a moment, I'll tell you the top five housing markets for AI-led home price growth, but how does AI investment push up home prices anyway? How does that happen? People are getting high salaries, signing bonuses, and stock options that produces well-funded buyers. They make big down payments, or they even pay all cash for homes, and when a buyer pays all cash for a home, they can pay absolutely any price because they don't have to get an appraisal that comes along with a loan for a financed property.   Keith Weinhold  17:53   That's how all cash buyers can really push up prices. The growth in AI companies that has really helped push the S and P 500 higher that fuels a wealth effect nationwide that makes everybody feel wealthier regardless of where you live as long as you're invested in the stock market but the localized effects with those higher AI wages and signing bonuses in order they are most potent in San Francisco, San Jose, Seattle, New York City, and Boston, and none of those are good cash flow investor markets. Still, short housing inventory is contributing to higher prices, and hey, it's time that we check on this again. Ever since the inventory crunch started to plummet in 2021 and reached its lowest point in 2022, I've been updating you on the housing supply, and I always keep it same same. I cite the same data source, the Federal Reserve Economic Data's active listing count, Fred's active listing count, which counts single-family and townhomes and condos, all wrapped up in this number. And the figure it still hasn't recovered at 1.1 million homes. Now it is 2% higher than last year, 2% more supply than last year, but overall housing supply is still 9% below pre-pandemic levels. And there's one important thing to keep in mind that most don't think about when you hear that figure that housing supply is 9% below pre-pandemic times in 2019, that does not mean we're 9% short. That is because even in 2019 there was a housing shortage, and we are 9% below that yet, keeping. Upward pressure on prices and the most supply-constrained markets today. It includes both good and poor cash-flowing investor markets.   Keith Weinhold  20:10   They are New York City, Chicago, San Francisco, Hartford, Providence, Milwaukee, Boston, Cleveland, Virginia Beach, and Kansas City. All of those places remain especially tight with housing inventory, and then finally, this fourth of four reasons I've cited for continued upward pressure on home prices are the fact that distressed sellers-they are few and far between-and you need a lot of those in order to have a serious down cycle, after the 2008 housing crash, millions of owners were underwater. They owed more on their homes than they were worth. Lending standards were irresponsibly loose. Adjustable rate mortgages were resetting higher. I mean, a lot of people had little choice but to sell or to hand the keys back to the bank. Distress, distress, distress. Today is almost the mirror image. Here's what's really happening with homeowners having this record equity position today-an average of over $300,000. Many also locked in at fixed mortgage rates below 5% it means that they're enjoying perhaps the cheapest long-term debt that they are ever going to have. Lending standards have been strong, foreclosure rates remain low, and virtually nobody is being forced to sell. That matters more than most people think because housing crashes need a lot of forced sellers, owners who must accept almost any price in order to escape the property. But today, most homeowners they can simply either stay put, or if they're going to move out of the home, keep it and rent out the home, or they can wait for a better offer. No distress. In other words, buyers might be frustrated, but sellers-they're just not desperate. And without desperation, it is difficult for home prices to fall sharply. So the bottom line here with today's home prices and looking into next year, home price growth is apparent, but it's weak. The ingredients for a national price collapse are nowhere to be found, so this does not spell boom or crash. Home prices appear poised to keep slowly grinding higher, but with this low affordability, that keeps them from soaring, say 10 or 12% higher. I don't see that happening. And of course, each December, I make my home price forecast to the exact percentage point for the year ahead, so you can look forward to that soon. The Get Rich Education home price appreciation forecast that I made late last year for this year. It looks like it's going to be almost spot on. Of course, unlike a lot of analysts, transparently, I also give you the result of how closely the forecast hit the target every year, so you can look forward to that too. Hey, if you like this show, there's more content where this comes from. Sign up for our complimentary newsletter. That way, you can see the graphs and charts and maps that I break down. If you like what you hear on Get Rich Education, every week I show you what's really happening with real estate rents, inflation, interest rates, and the economy, and more importantly, what you can do about it. You'll get sharp insights, useful opportunities, and a few laughs along the way. Yeah, a couple knee slappers sprinkled in there with actionable strategies, like the savviest way to get rent increases. Get smarter in just a three to four minute read every week. Join 1000s of smart investors right now at greletter.com because your inbox could use fewer coupons and more financial freedom. That is greletter.com. More straight ahead.   Keith Weinhold  24:20   I'm Keith Weinhold. You're listening to Get Rich Education. What if you got your mortgage loans the same place I get mine? You sure can at Ridge Lending Group NMLS 42056. They provided GRE listeners with more loans than anyone because Ridge specializes in investment property. They'll help you build a long-term plan for growing your real estate empire with leverage. Start your prequal and even chat directly with President Chaley Ridge. While it's on your mind, start at ridgelendinggroup.com. That's ridgelendinggroup.com. Let me ask you something. If you've worked hard to build wealth, is your. Money positioned to actually support your goals. A lot of accredited investors leave capital sitting in cash because it feels safe, but inflation and missed income opportunities can quietly erode its value. Freedom Family Investments offers freedom notes for investors seeking structured income backed by real estate. It's a straightforward approach built on real assets, not speculation. In full disclosure, I'm an investor myself. What I like is that their team walks you through how it all works, so you can decide if it aligns with your portfolio and income goals. Every investment carries risk, and nothing is guaranteed. But with a track record of consistent, on-time investor payouts. They built real credibility. Go to freedomfamilyinvestments.com to book a clarity call, or text family to 66866. That's family to 66866.    Dana Dunford  25:59   This is Hemline's co-founder Dana Dunford. Listen to Get Rich Education with Keith Weinhold, and don't quit your daydream.   Keith Weinhold  26:15   Welcome back to Get Rich Education. I'm your host Keith Weinhold. There will only ever be one episode 622, and you're listening to it. I hope you're enjoying the late summer. I'm wringing every bit of time and enjoyment out of it that I can. I don't know if this part was enjoyable, but I ran an all-out mile on a track. I wanted to see how fast I could run a mile. I had a friend pace me, and I got a 631. I was happy with that since I hadn't done any specific training. Yes, a mile is more than four laps on a track as well. Did you know that? Yes, this detail-oriented shaved mammal here diligently measured off that extra nine point something meters. Ah, I'll tell you that fourth lap hurt so badly that if my buddy weren't there, I might have just quit and not finished the mile. But summer's days are numbered, and that's too bad because it is my favorite season of the year. The NFL season kicks off in just two days on the ninth, with Seattle hosting the New England Patriots in a rematch of last year's Super Bowl. So then, I guess it looks like your productivity for the week will end with a respectable two-day run as you tune in to that game. Where is the future demand for real estate going to come from? It comes from a growing population. The U.S. is expected to add 21 and a half million people from 2025 to 2040. 21 and a half million more people. The overall population it's expected to grow from about 341 million up to 363 million. That is where we're going. That's per the Census Bureau and the University of Virginia, projecting 341 up to 363 by the year 2040, which is just a little over 13 years away. Okay, so that part is not so surprising, but here is what is absolutely staggering: more than half of this entire increase is projected to occur in just two states, just two of the 50 states, more than half of the increase. Do you know what they are? In fact, I showed you a map of this in a recent newsletter, but I can talk about it and expand on it more here.   Keith Weinhold  28:52   The two states that are expected to account for more than half of the nation's overall population growth through 2040 are Texas and Florida. They're already the second and third most populous states, respectively. It's kind of like America looked at the map, checked their weather app, and started packing sunscreen. Texas is expected to add 6.6 million residents. Florida welcoming another 4.6 million during this span. So that is over 11 million new people between them. This is like taking the entire population of Georgia and dropping it into those two already booming states, that much growth in this fairly short period of time, for real estate investors, more people that generally means more demand for our housing product, and I'll get back to the staggering Texas and Florida imbalance in just a moment. Because there are big gains in other investor-friendly southeastern states like Georgia and Tennessee, the Mountain West should swell alone. The South, okay, the region that the Census Bureau delineates as the South, which sort of runs from Maryland all the way down south and then west out toward Texas, the South just until 2040 is expected to account for 78 percent of the growth. That is just staggering. Cash flow hotbed Indiana that should grow by nearly a quarter million residents as well. The Carolinas are ballooning. Already the most densely populated state in the nation, New Jersey, that will get more dense with some pretty healthy population growth. Its residents have not discovered elbow room, but not every state is adding population. 14 states are expected to shrink, led by Illinois losing 650,000 people and New York down 457k. Again, this is all through 2040. In fact, a small loss cluster actually runs through the South, though West Virginia, Mississippi, and Louisiana-they're projected to lose 440,000 people combined. You know that whole theory that sometimes you hear people talk about, like with Earth warming and drying, you're going to have people stampeding toward the freshwater Great Lakes states. That is probably farcical. That just has not shown up in the data. That people are moving in droves to say cooler Michigan and Wisconsin for those reasons.   Keith Weinhold  31:46   It's just not happening now. Of course, population projections are not delivered from Mount Sinai on stone tablets. Besides births and deaths, the level of future immigration, of course, that's the real wild card here. After the Trump presidency ends by 2029, the next administration that could tighten or loosen the immigration spigot, that could materially reshape the map. But they're probably not going to tighten immigration. I mean, they couldn't because the flow really couldn't be crimped much more than it already is. People love to poke fun at California, but even in 2040, it is expected to barely retain its crown and edge out Texas to still be the most populous state: 39 million versus 38 million, respectively, for California and Texas by 2040. But yeah, Texas and Florida-they are the real stories here, and why droves of people are attracted there for cheaper housing, jobs, warm weather, a business-friendly environment, and Texas and Florida are also places where builders can still build without completing some side quest worthy of a video game with all their permits and regulations and roadblocks. You're largely free of those things in Texas and Florida. Now there are two more important factors to keep in mind here. Some bigger picture context. I've talked before about how the overall American mobility rate is down, and this is a long, long trend. Decade after decade, fewer people move and more people stay put, which is contrary to popular belief. This lower mobility rate, and another factor that gives you perspective is that as real estate investors, we know all this stuff I've been talking about here. These population changes-they only look at the demand side. The supply side matters just as much, despite their slower population growth. Northeast and Midwest states build less new inventory, and that is why Northeastern and Midwestern housing prices and rents are still growing faster today than they are in the Sun Belt, despite all of those Sun Belt construction cranes. You know, too many construction cranes. It looks bullish, and it actually is, but it spikes supply and it suppresses prices. And really, the bottom line here with American population growth from now until 2040 is follow the people, but count the rooftops. Population growth creates housing demand, while limited construction creates scarcity.   Keith Weinhold  34:46   The best opportunities often emerge where those two forces collide. That's what you really want to look for: demand and scarcity. Now, the apartment space. We all know that's been beleaguered for about three or four years, ever since higher mortgage rates set in and high construction levels conspired to keep apartment rents suppressed. In fact, multifamily construction had a peak in this cycle during 2024. That's when 600,000 units were built back in 2024. That was the most new apartment supply since 1986. That is when Cheers, MacGyver, and Miami Vice were on television. Run DMC was on urban radio. MTV was a dominant cultural force, the most new apartment supply since 1986. That's when kids were playing with GI Joe's, He-Man, and My Little Pony. For adults, fashion-wise, they were wearing enough shoulder padding to survive a minor collision. So, lots of new apartment supply to get absorbed. It is getting more and more absorbed. There are more signs there now because the national median apartment rent has now increased for seven months in a row. That's according to Apartment List. Also, the apartment vacancy rate has dropped for six straight months, and do you have any idea what the national apartment vacancy rate is? It has dropped down to now 7.1% Inevitably, overbuilt apartments will be absorbed with a growing population. Lots of great episodes coming up here on the show, where you might be in for a surprise next week. A renowned macro economist will be here on the show with us. I think we all know that in 1971, the U.S. had a lot of economic changes. That's when Nixon completely eliminated us from the gold standard, and the economic system shifted from capitalism to creditism back then. Well, now we appear to be leaving creditism and entering a new economic phase. This could be seismic. Next week here on the show, he'll reveal what the new era is called and how you need to prepare for it, that's next week here on episode 623. If you haven't yet, be sure to hit the follow button or subscribe button on your podcatcher so that you don't miss it.   Keith Weinhold  37:31   Again, if you like what you hear here each week, the GRE "Don't Quit Your Daydream" letter gives you the sharpest ideas of the week in about three or four quick hitting minutes, you'll get surprising housing data, wealth building strategies, timely opportunities, news that a lot of times you can't get anywhere else, and maps and charts that make you say, "Wait, what? It's smart, useful, entertaining, and completely free. Thousands of investors read it every week, and believe it or not, I'm actually more of a writer than a talker. Don't just listen to Get Rich Education, get the letter at greletter.com. That's greletter.com. Until next week, I'm your host Keith Weinhold. Don't quit your daydream.   Speaker 2  38:23   Nothing on this show should be considered specific, personal, or professional advice. Please consult an appropriate tax, legal, real estate, financial, or business professional for individualized advice. Opinions of guests are their own. Information is not guaranteed. All investment strategies have the potential for profit or loss. The host is operating on behalf of Get Rich Education LLC exclusively.    Keith Weinhold  38:51   The preceding program was brought to you by your home for wealth building, getricheducation.com

Send us Fan MailFind Elaine on Instagram at PageantpreneurA listener note before you press play. This conversation includes grooming by a church leader, an eating disorder, and drug addiction.Some stories are not for scrolling past. This one belongs to Elaine Mingus.Her father moved the family out of Houston to keep his kids away from the big city. They started going to church in a small Texas town. The youth pastor there was grooming Elaine and several other girls in the group.When it came out, the family sat down with the senior pastor. Before they said a word, he told them he already knew. The youth pastor had confessed. And the church believed Elaine had seduced him.She was a teenager.This was the nineties. Nobody used the word grooming. Nobody suggested therapy. So she buried it, and she carried the shame instead. That shame turned into eight years of bulimia and an almost daily heroin addiction.Then she walked into a church, saw the crucifix, and says every question she had was answered right there, including ones she did not know she was asking.She is careful to say the story is not finished. Jesus is not done yet.Today Elaine is Mrs. Texas, a mother of seven, and the founder of Pageantpreneur, where she helps women in pageantry use their story, voice, and platform to get visible, land sponsors, and get media attention that outlasts the crown.The conversation she is most interested in is not how to become successful. It is what happens after you spend years believing the next accomplishment will finally make you feel like you have arrived.She has a line for it. Yesterday's success becomes today's baseline. So the goalpost moves and moves and you never once stop to look back. She thinks that is why God kept telling the Israelites to build altars.She also talks about counting. Calories, the scale, view counts, hours with your kids versus hours without them. Always calculating your worth from your own performance instead of receiving what Jesus already settled.There is a working idea in this episode you can use today. Elaine builds every client around one overarching story and five signature stories underneath it. Then everything she posts checks back against those six. She says women worry they are repeating themselves, but people want predictability. It is what makes them trust you.Her other line worth keeping. You are not responsible for the outcome. You are responsible for the action.Find Elaine on Instagram at Pageantpreneur for the business, or at https://pageantpreneur.comThis is The Calling Table Sessions on the Stuck No More Voices Podcast. One seat. One woman. One story handled with care.If you are a Christian woman leader with a story that could reach the woman you used to be, there is a seat here for you at theresacroft.com/yourseat.Share this episode with the woman who came to mind while you were listening.Subscribe on YouTube. Share on Apple Podcasts or Spotify.Hosted by Theresa Croft, a Visibility Strategist who helps Christian women move from their calling to their story to being found with a message for their ideal audience.Register Click HERE:  The Calling Activation Study; Discover the Story That Connects Your Calling to the People You're Meant to Serve.To be a guest on the Calling series:  https://theresacroft.com/your-seatHome TheresaCroft.comInstagram: https://instagram.com/theresacroftFacebook: https://Facebook.com/theresamcroftYouTube: https://YouTube.com/@theresacroftMore Podcast Episodes on Apple and Spotify

Week One is Here! Josh and I are back to breakdown the first true week of college football. We talk about the exciting matchups and what to expect from week one. Plus, a few other thoughts along the way.

Send us Fan MailResources RECOVER Diagnostic: https://eligibility.natrevmd.com/recover-quiz-lp natrevmd.com Payment Posting Audit Checklist: https://eligibility.natrevmd.com/payment-posting-checklist Show notes A provider sees a patient for a scheduled procedure. A separate problem comes up mid-visit, gets evaluated, and the practice bills both services with Modifier 25 attached. The claim pays, and everyone moves on, until that same claim gets swept into a targeted payer audit because the documentation never actually supported a separate, significant E/M service. The three failure patterns Routine pre-procedure work billed as a separate visit:  Baseline assessment before a procedure, confirming the patient is appropriate, reviewing labs, checking vitals, is part of the procedure. It is not a separate E/M, and Modifier 25 does not apply just because something happened before the procedure. Cloned or thin documentation:  An assessment and plan identical to the note from two visits ago, or a problem mentioned in one line with no distinct plan, will not survive a payer review. The documentation has to show medical decision-making distinct from, and above and beyond, the procedure. Modifier 25 used to override a denial:  A claim gets bundled and denied, someone appends Modifier 25 and resubmits, and it pays. If the documentation never supported a separate E/M, that resubmission was not a correction. It was a workaround, and it is exactly the pattern payer audits look for.The global period trap Global periods run 10 days for minor procedures and 90 days for major surgeries. During that window, routine E/M care for the same procedure is bundled and not separately billable, even with Modifier 25 attached. A genuinely unrelated new problem may have a path to separate billing, but it needs documentation of the unrelatedness and compliance with payer-specific global period rules. Procedural specialties, surgical groups, orthopedics, gastroenterology, dermatology, OB/GYN, carry the most risk here. The five-question audit test 1.  Was a significant E/M service actually performed, beyond the usual work of the procedure? 2.  Is the separate problem, assessment, and management clearly visible in the documentation? 3.  Would the E/M have been separately reportable if the procedure had not occurred that day? 4.  Do current NCCI, global-period, and payer-specific rules allow Modifier 25 here? 5.  Could the practice defend this claim on the medical record alone, not just the modifier? Three actions this week Pull 20 to 30 Modifier 25 claims from the last 90 days across your most frequent providers and run each one through the five-question test. Where claims fail, start with provider education, one conversation with examples from their own documentation, not a policy memo. If more than 20 percent of the sample fails, add a pre-release review for high-frequency or high-risk providers for 60 to 90 days while the pattern corrects. Episode breakdown The setup: what Modifier 25 is actually supposed to communicate The three failure patterns Three cases: yes, no, or verify The global period trap The five-question audit test Running your own Modifier 25 practice audit 

Thank you to The Commons for supporting this episode: https://www.thecommons.com.au/The biggest stories on the internet from September 3rd, 2026.Please consider buying us a coffee or subscribing to a membership to help keep Centennial World's weekly podcasts going! Every single dollar goes back into this business

We've got a repeat guest on the show today. Alex Dow is back. Co-founder of Mirai Security, and currently an enterprise security architect for a large financial firm, and honestly one of my favorite people to talk security with. Alex has done federal security work, and securing the actual Olympics, now then a security consulting firm.If you've been listening a while, you might remember Alex from way back in episode 11 and then again in episode 96.Today's round three, and we're going fully nerd-out mode on the OpenAI/Hugging Face hack. Walking through the signals and understanding how this type of attack was seen from a defensive side.The timeline visualization of the Hugging Face attack that Alex referenced.This episode is brought to you by Opsleader Pro. A place for MSP owners and managers to get the systems and tools they need to build a stable and growing MSP. Part group coaching, part peer group, everything you need to run a successful MSP. (00:00) - Welcome Back Alex Dow (00:50) - What Happened in the Hack (02:22) - Mythos and AI Hype (06:05) - Throughput Beats Creativity (08:04) - Sandbox Escape Explained (11:02) - Swarm Agents and Breadcrumbs (14:13) - Reward Hacking and Alignment (19:30) - Four Days Undetected (21:27) - Blue Team Limits and Guardrails (25:28) - Detecting AI in the Noise (30:50) - Using Your AI Against You (34:30) - Honeytokens and Deception (37:22) - Zero Trust as a Baseline (38:59) - AI Botnets and Key Theft (43:01) - Wrap Up and Takeaways

Max Brending und Jonathan sprechen über aktuelle Seahawks-News, ordnen die Fantasy-Optionen des Teams für die kommende Saison ein und geben kurze Draft-Tipps. Außerdem nennen sie ihre „MyGuys“ und „Not-MyGuys“. Kapitel 00:00:10 – Fantasy-Football bei den Seahawks 00:05:07 – Der Draft startet jetzt 00:09:53 – Seahawks-Spieler im Fantasy-Check 00:17:08 – Receiver mit und ohne Value 00:24:43 – Running Backs und Handcuffs 00:33:20 – Tight Ends und späte Targets 00:37:55 – Quarterback-MyGuys im Überblick 00:45:02 – Meine liebsten Receiver-Ziele 00:52:52 – Michael Wilson als Breakout-Kandidat 00:57:26 – Boston als Spätjoker 01:02:44 – Running-Back-Debatte beginnt 01:09:07 – Baseline statt Risiko 01:11:17 – Späte Running-Back-Perlen 01:13:29 – Tight-End-Targets in der Breite 01:17:45 – Quarterback-Risiken 01:23:42 – Receiver zum Meiden 01:33:06 – Laufspiel mit Fragezeichen 01:37:47 – Tight Ends und Fades 01:41:41 – Fantasy-Regeln für den Draft Der Ballhawks-Podcast der German Sea Hawkers begleitet die Seattle Seahawks seit 2015 mit Analysen, Diskussionen und Neuigkeiten rund um Team, NFL und Fanklub. Alle Episoden und das vollständige Archiv: https://www.germanseahawkers.com/podcast-index/ Spotify: https://open.spotify.com/show/3nZOfYlPMvmTZTEvt4bhHs Apple Podcasts: https://podcasts.apple.com/podcast/id1033877656 Amazon Music: https://music.amazon.de/podcasts/d7bb4104-4e94-4dae-88e1-8067a44abd86/ballhawks--podcast-der-german-sea-hawkers-ev Deezer: https://deezer.page.link/LbEHV27kEEVM7d9T9 YouTube: https://www.youtube.com/@GermanSeahawkers/ Fragen, Anregungen und Kritik: podcast@germanseahawkers.com Go Hawks!

Football is back! Josh and I discuss week 0, plus we discuss our preseason power rankings and heisman watch list. This is our favorite time of year, but we also need to talk about the Browns terrible move of starting Watson.

In episode 387 of The Physical Performance Show, host Tim Studley picks up part two of his conversation with fellow Pogo Physio physiotherapist Lewis Craig, continuing last week's deep dive into surf life saving and Ironman racing. This episode shifts focus from injury patterns to managing niggles, optimising performance, and the habits that separate top-tier surf ironman athletes from the rest. Tim and Lewis discuss the collaborative decision-making involved in managing a recurring niggle, the key training and screening recommendations every surf life saving athlete should have in place, the lessons Lewis has learned working with elite athletes in the sport, and his top tips for improving performance — from sand-specific training to fuelling properly for a demanding training load. SHOW SPONSORS:

Josh and I celebrate 250 episodes of the podcast by talking about football. What else would we talk about? We breakdown the last two divisions in the NFL and of course talk about Notre Dame.

Josh and I are back at it again with another football preview! This time, we discuss the NFC North and AFC South! One very good division and one not so good division. Who do you think pulls it out this year?!

On this episode of Fall Obsession Podcast, Sam sits down with Casey Chamberlin from Baseline Maps, one of the rapidly growing mapping apps making waves in the fishing and hunting community. Casey shares the story behind Baseline Maps, how the app has grown, and what makes the community surrounding it so unique. They dive into the vision behind the platform, the constant evolution of the app, and Casey's hands-on approach to listening to hunters, responding to feedback, and building Baseline around the people who actually use it in the field. They also talk about the role technology is playing in modern hunting, how Baseline is helping fishermen and hunters scout and understand the ground they hunt and fish, and why building a strong community is just as important as building a great product. From the early days of Baseline to where the app is headed next, Casey gives us an inside look at what it takes to grow an outdoor technology company while staying connected to the hunters who helped build it. The app is growing. The community is growing. And Baseline Maps is just getting started!Fall Obsession Podcast is sponsored by:Hoot Camo Company (https://hootcamo.com/) - use code "fallobsession15" to save with HootBear River Archery (https://www.bearriverarchery.com/) - use code "fallobsession" when shopping online with Bear RiverTactacam Reveal Cameras (https://www.tactacam.com/)The Outdoor Call Radio App (https://www.theoutdoorcallradio.com/)

Covert narcissism doesn't just create crisis moments — it leaves your nervous system stuck in a chronic state of hypervigilance long after the argument ends. You can get good at managing the spike, catching yourself mid-reaction, and still lie awake at night with your jaw clenched and your shoulders up around your ears for no reason at all. In this episode on narcissistic abuse recovery, Renee explains allostatic load — what happens when a nervous system adapts to years of walking on eggshells, gaslighting, and unpredictable moods in a covert narcissistic relationship, and never fully comes back down. She breaks down what chronic hypervigilance actually looks like from the inside, what it does to the body physically, and introduces eight new tools for lowering your baseline over time: Worry Spot, Villain Voice, Letter to Future You, The Doorway Reset, The Off-Duty Ritual, Compliment Jar, The Do-Nothing Timer, and The Petty Playlist. This is part two of a two-part series on covert narcissism and fight or flight. If you haven't heard part one, start there for eight in-the-moment crisis tools built for the spike itself. If you've been listening, learning, and starting to see the patterns of narcissistic abuse in your relationship — but still feel stuck in the same cycles of emotional dysregulation and self-doubt — that's not because you're missing information. It's because this isn't something you're meant to untangle alone. If you're ready to take the next step toward healing, I invite you to check out my coaching program at www.covertnarcissism.com. And if this episode helped, please subscribe so you don't miss what's next. The information provided by Renee Swanson, Covert Narcissism Podcast, and CNG Life Coaching is for educational purposes only and is not to be used for diagnosis purposes and is not intended to be a substitute for clinical care. Please consult a health care provider for guidance specific to your case. This material discusses narcissism in general. Renee shares stories from her personal experiences as well as from those she has talked with for several years. Her material does not claim that any specific person has narcissism and should not be used to refer to any specific person as having narcissism. Permission is not granted to link to or repost this material to support an allegation or support a claim that any specific person is a narcissist. That would be an unauthorized misuse of the material and information provided. Hosted by Simplecast, an AdsWizz company. See pcm.adswizz.com for information about our collection and use of personal data for advertising.

I morgen kl. 16.31 ruller Matson ned ad rampen i Monaco og starter sin anden Vuelta a España i karrieren. Det er fire år siden, han sidst stod på startstregen i Vueltaen, men det føles næsten som en helt ny debut. Hør hvorfor, og få alle indtrykkene fra Monaco, Golden Tulip i Frankrig og optakten til La Vuelta a España 2026. Medvirkende: Anders Mielke & Mathias Sunekær Norsgaard.  Gruppettoen på Forhjulslir er sponsoreret af Aioss.  Som ny kunde hos Aioss i august måned modtager du en "Back to Baseline"-gave – du modtager kreatin svarende til 6 måneders forbrug. Ved at købe Aioss støtter du ikke bare os – men vigtigst af alt dig selv, med mere fysisk og mentalt overskud i hverdagen. Brug koden "gruppettoen" og spar 100 kr på de tre første leveringer på dit aioss-abonnement. Læs mere på: https://aioss.dk/pages/gruppettoen

Jesus Water at the park. More than Waffles and Credere's dislike for breakfast cuisine. Baby toys scare baked Brandon at night. Baseline for bullshit is at zero now. We then give you 40 mins of NFL 2026 season preview ripe with division winners and stories for each. MVP dark horses and coach of the year. Penguins are like Bun B on this weeks WASSUP (weird animal segment).

Gruppettoen optager i hver deres fyrstendømme - Mielke er på Bornholm med svigerfamilien og Matson er i Monaco, hvor han forsøger at trænee, kæmpe og svede sig til en plads på Vuelta a España-mandskabet. Hør mere om den varme kamp, Mielkes bornholmske eventyr og meget mere.  Gruppettoen på Forhjulslir er sponsoreret af Aioss.  Som ny kunde hos Aioss modtager du en "Back to Baseline"-gave – du modtager kreatin svarende til 6 måneders forbrug. Ved at købe Aioss støtter du ikke bare os – men vigtigst af alt dig selv, med mere fysisk og mentalt overskud i hverdagen. Brug koden "gruppettoen" og spar 100 kr på de tre første leveringer på dit aioss-abonnement. Læs mere på: https://aioss.dk/pages/gruppettoen

If you find yourself constantly resetting back to square one, you aren't building capacity—you are managing symptoms.True resilience isn't defined by how fast you bounce back from pressure; it is measured by how steady your baseline remains when pressure hits. Every experience creates a somatic-mindset feedback loop. When you autonomously self-regulate, either you leverage these moments to expand your baseline capacity, or you stagnate in a cycle of temporary recovery that perpetuates as is over time.In this episode, we unpack why conscious leaders, spiritual practitioners, and self-led individuals utilize somatic-mindset micro-habits to transcend the reset loop and move into an enlightening and intelligence growing loop instead. Discover how shifting from basic emotional control to mastery and dynamic self-leadership allows you to expand your capacity and navigate pressure from a grounded baseline. This foundation creates empowered composure, allowing you to uphold relational integrity and co-regulation.In This Episode, We Cover: The Symptom Management Trap: Why relying on repeated "resets" keeps your baseline static. The SMART MICRO-HABIT Somatic-Mindset Loop: How real-time somatic awareness and conscious mindset choices strengthen your baseline under pressure. Expanding Capacity vs. Managing Activation: Transitioning from reactive coping mechanisms to proactive baseline expansion. Somatic-Mindset Micro-Habits in Action: Practical, high-impact strategies to maintain emotional sovereignty and personal accountability.Episode Timestamps:(0:00) — Beyond Symptom Management: Somatic-Mindset Micro-Habits & Expanding Capacity (Window of Tolerance to Welcome)(5:10) — Emotional Sovereignty & Self-Leadership: Evolving from Control to Dynamic Regulation(7:14) — The Liberating Shift: Empowering Personal Sovereignty through Somatic Integration(7:40) — Recommended Reading for Nervous System Sovereignty & Mindset Mastery(7:59) — Moving Beyond the Reset: Aligning Intrinsic Motivation with Adaptive Self-LeadershipResources & ConnectSubscribe & Share: If you know spiritual and conscious, regulated leaders who want to navigate reactive survival loops with self-compassionate honesty and steady empowering composure using somatic-mindset micro-habits share this episode with them.Explore the Frameworks and digital resources on the Payhip store: Access dedicated somatic-mindset tools, guides, and workbooks designed to support baseline expansion: https://payhip.com/InspiringHumanPotentialShift from Reactive Survival Loops to Real-Time Empowered Composure—The SMART Way. Download Your FREE 10-Minute SMART MICRO-HABIT Check (PDF): https://payhip.com/b/r3sNW

Journalist. Podcaster. Network founder. Warren Shaw joins Kevin L. Warren to explain how 19 Media Group and The Baseline NBA Podcast were actually built — cold calls, credential hustles, and a COVID-era decision that changed everything.

GLP-1 medications have changed how we treat obesity and metabolic disease. But as their use has exploded, so have questions about side effects, muscle loss, long-term use, and whether patients are receiving the support they need to use them safely. In this episode, I reconnect with metabolic health and regenerative medicine expert Dr. Tyna Moore to revisit our conversation from two years ago and examine what we've learned since. We discuss: How to tell when your GLP-1 dose may be too high What you can do to protect your muscle and bone during weight loss Which metabolic and nutritional markers should you check before and during treatment Why weight can sometimes return after stopping a GLP-1 What emerging research suggests about GLP-1s beyond weight loss GLP-1s can be life-changing, but a lower number on the scale isn't the same as better health. Ultimately, how these medications are used—from dosing and monitoring to nutrition and strength training—matters just as much as whether they're used at all. Additional resources: Join Dr. Tyna Moore's community Listen to Dr. Tyna Moore's previous appearance on The Dr. Hyman Show View Show Notes From This Episode Sign up for Dr. Hyman's Brainshaping Academy to learn how to nourish the biological systems that support your mental, emotional, and cognitive health https://drhyman.com/products/brainshaping?utm_source=dr_hyman_show&utm_medium=newsletter&utm_campaign=may_27&utm_content=link Get Free Weekly Health Tips from Dr. Hymanhttps://drhyman.com/pages/picks?utm_campaign=shownotes&utm_medium=banner&utm_source=podcast Sign Up for Dr. Hyman's Weekly Longevity Journalhttps://drhyman.com/pages/longevity?utm_campaign=shownotes&utm_medium=banner&utm_source=podcast Join the 10-Day Detox to Reset Your Healthhttps://drhyman.com/pages/10-day-detox Join the Hyman Hive for Expert Support and Real Resultshttps://drhyman.com/pages/hyman-hive This episode is brought to you by Seatopia, Perfect Amino, Cozy Earth, Timeline, Sunlighten, and Made In. Find a cleaner source of seafood. Check out seatopia.fish and use code HYMAN for free shipping on your first order. Get daily protein support at bodyhealth.com and use code HYMAN20 for 20% off. Head over to cozyearth.com to save 20% and upgrade all of your daily essentials today. Support healthy aging and get 20% at timeline.com/drhyman with code HYMAN. Discover why so many people are using sunlighten.com and use code HYMAN to save up to $2,100 today with free shipping. Upgrade your cookware at madeincookware.com and save 10% off your first order with code HYMAN-HIVE. (0:00) Introduction, Dr. Hyman's evolving views, and episode goals (0:43) Sponsor: Rose Nutrition Liposomal NAD (1:42) Sponsor: Seatopia clean seafood box (2:44) Disclaimers and Lyme disease preview (4:04) Guest Dr. Tina Moore reintroduced (4:30) GLP-1s: Effects after years and microdosing strategies (7:14) Risks of high-dose GLP-1s and misconceptions about muscle/bone loss (13:09) Functional deficiencies and microdosing approaches (17:05) Sponsor: Made In stainless clad cookware (18:02) Sponsor: Timeline with Mitopure (18:58) Broader and additional benefits of metabolic health and GLP-1s (21:48) GLP-1s for immune and brain health; genetic differences (27:59) Introduction to peptides and GLP-1 drugs (32:36) Gray market concerns and weight regain after stopping GLP-1s (37:14) Long-term safety, cost, and personalizing GLP-1 treatment (40:57) Emotional blunting and recent concerns about GLP-1s (46:30) Functional medicine approach: addressing root causes (46:46) Sponsor: Sunlighten Sauna (47:20) Sponsor: Magnesium Breakthrough from Bio Optimizers (48:17) Dr. Hyman's evolving perspective on GLP-1s (49:19) Hormonal effects of GLP-1s for men and women (55:02) Baseline lab markers and tests before GLP-1s (57:10) New and next-gen GLP-1 therapies (59:43) Telemedicine, gray market issues, and importance of reputable practitioners (1:05:57) Rapid fire: Alcohol, common mistakes, misconceptions, and eligibility for GLP-1s (1:08:16) Key lab tests and surprising non-weight benefits

After a week off, Josh and I are back with our football previews! We break down the Big Ten and the AFC North. Who will come out on top of the Big Ten? Will Indiana do it again or will Oregon come out on top? Then do the Browns have any chance to win the AFC North?

You wake up already reactive, responding to someone else's needs before you've taken a single conscious breath. And by 8am you're somehow already spent, already behind.That's not a discipline problem. The first thirty to sixty minutes of your day quietly set the cortisol curve and stress baseline you'll run on for the next twelve hours, whether you choose it or not.This week's Hi-Cap Move reframes the morning routine as nervous system strategy, not a productivity hack, and gives you five simple principles (no rigid 5am protocol) to start the day in regulation instead of survival.If your mornings currently belong to chaos and you're tired of being along for the ride, this is where you take that window back.--

Build Your Base - Lifting Plan to Build Capacity Needed for PoleConfusion to Clarity - Masterclass on How to Juggle All The ThingsInstructor Course Interest listAll my offers/services➡️ Online Summit that I'm a speaker inConnect with Dr. Emily:Website - i have tons of resources for you!Instagram

One of the greatest examples of basketball's "Special Teams" is your baseline out of bounds philosophy and execution on both sides of the ball. In today's podcast, we discuss goals that you should have for your playbook, who the ball should go to, formations, counters and selling the importance of baseline out of bounds to your team! We also have a great quote from USA Basketball, and we also discuss today in our Coaches Study one of my friends in the business who made his name nationally via USA Basketball!

Global water systems are facing an unprecedented challenge as climate change, resource depletion, and globalization place staggering pressure on our natural resources. Dr. Raha Hakimdavar, Senior Advisor to the Deans of Georgetown University in Qatar (GU-Q) and the Earth Commons & Founder & CEO - Zyon Space, unpacks the critical intersections of hydrology, space science, and environmental security. From analyzing how agricultural water waste and food imports impact the Arab world to detailing the vulnerabilities of desalination infrastructure, she offers a profound look at the cascading threats facing urban centers. By leveraging advanced satellite data collected from space, she reveals how tracking global groundwater levels can transform these invisible vulnerabilities into vital opportunities for international cooperation and policy reform.   00:00 Introduction 04:36 Energy, Oceans & the Cost of Innovation 06:22 Supermarket Aisles & Disconnected Systems 10:09 A Fifty-Year View from the Cosmos 14:09 Urban Budgets & Changing Taps 26:02 The Changing Anatomy of Regional Diets 30:47 Invisible Pipelines & Cascading Points 34:58 Distorting the Baseline of National Security 36:14 The Enabling Elements of Uprisings 38:06 A Landscape of Treaties   Dr. Raha Hakimdavar is a hydrologist, science policy expert, and space science leader with a proven record of innovation across government, academia, and industry. She is currently a Research Professor at the Earth Commons and Senior Advisor to the Deans at Georgetown University in Qatar and the Earth Commons, leading strategy and programs focused on environmental security, climate action, and sustainability research and education. She has active research projects in Greece (water scarcity), the Middle East (water and food security), and Indonesia (flood risk and nature based solutions), with a special focus on small islands. Dr. Hakimdavar has served as a technical consultant for UN Environment, the World Bank, and USAID on disaster risk reduction, water, and forestry projects for over a decade. Dr. Hakimdavar is also the Founder and CEO of Zyon Space, which supports emerging space agencies and organizations in the Global South in leveraging Earth observation technologies for climate and environmental resilience.   Connect with Raha Hakimdavar

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

Episode 81 of the Unspoken Words podcast features Dr. Elisa Shipon-Blum and Dr. Jenna Blum returning to the foundation of it all — what S-CAT, or Social Communication Anxiety Treatment, really is, and just as importantly, what it's not. Prompted by families who reached out to the SMart Center still confused about the approach, Dr. E and Dr. Jenna make the case that the question so many parents ask — "How do I get my child to talk?" — is the wrong one. The better question is why is my child not communicating in this situation, because not speaking is only a symptom of something deeper.At the heart of the episode is a simple truth: this is a whole-person approach, not a Band-Aid for a single behavior. Using Dr. Jenna's "cake" analogy, they explain that every child is made of different ingredients — and that missing even one can mean missing a pivotal piece of the picture. That's why selective mutism is best understood not by its stigmatizing name, but as a social communication anxiety disorder with underlying "whys," and why the SM Evaluation is the essential first step in ruling those whys in or out.Dr. E and Dr. Jenna explain why S-CAT is a recipe of evidence-based approaches rather than any single therapy — blending CBT, behavioral exposures, motivational interviewing, insight-oriented work, and intensive parent management. They walk through the baseline stages of the Social Communication Bridge®, the feelings chart and the "sweet spot" that keeps kids growing without tipping into avoidance, and the parent's role as the driver of real progress. Along the way, real stories bring it to life — from a teen learning to get his own tickets at a Phillies game to a nine-year-old who needed structure long after her mutism resolved.The episode closes on a powerful reminder: the longer selective mutism lingers, the more likely it is that a "why" has been missed — and that with the right understanding, no one has to remain selectively mute.--Chapters: (03:20) Why "How Do I Get My Child to Talk?" Is the Wrong Question—And What to Ask Instead(10:33) The Bridge, the Baseline, and Why SCAT Is a Recipe—Not a Single Strategy(14:37) Honoring a Child's Feelings and the Parent's Role in Driving Real Progress(26:26) Motivation, the Hidden "Whys," and Why Courage Isn't Always Loud(34:16) The Whole Person Behind the Symptom—And Carrying the Gains Beyond SM- ADDITIONAL RESOURCES: https://selectivemutismcenter.org/resources/ Ask Dr. E a question of your own! Learn more about the host, Dr. Elisa Shipon-Blum Explore our SMart Center success stories! Get started at the SMart Center Listen to other Unspoken Words episodes here. For the best clips from every episode, follow the podcast on Instagram & YouTube Learn more about CommuniCamp, our 3+ day intensive group treatment and ALL DAY parent training & support programLearn more about our 6-week, virtual social skills series, which are skills-based groups designed to help children, teens, & young adults build social communication, comfort, and connection with similar aged-peers in a supportive setting.- For all podcast inquiries, please contact Dakota Hornak at ⁠dhornak@selectivemutismcenter.org⁠ This podcast was produced and published by New Edition Productions (neweditionconsulting.com)

On this Iron Sights episode recorded live on the vendor floor at TTPOA Conference, Scott Howell sits down for a collab with the TTPOA Podcast hosts Brandon Hernandez and Matt Smith. Their guests are Brad Ortiz, Director of Sales for the Law Enforcement Division at Silencer Shop and former officer, and Dewayne Manson, Project Manager at the American Warrior Association, to dig into the thing the Law Enforcement community has been slow to confront: the compounding weight of moral injury, and why the reactive system built to support officers is failing the people who need it most.The conversation goes deep on what moral injury actually is, how it differs from PTSD, and why the officer who has never been in a critical incident can be just as broken as the one who has. Brad and Dewayne walk through what a culturally competent clinician actually looks like, why baseline brain and blood health mapping belongs in the academy, and how the American Warrior Association's R3 program is embedding proactive resilience training directly into TTPOA's regional structure, the first partnership of its kind nationally.But this episode isn't just about mental health. It's about identity. What happens to the guy who IS the job when the job ends? Brad gets personal about his own struggles stepping away from full-time law enforcement, Dewayne shares the moment his teenage son had to pick him up after a DUI, and Scott connects it all to the broader truth that your calling and your identity are not the same thing, and that confusing the two is costing officers their health, their relationships, and sometimes their lives before and after retirement.The back half gets into the business world and how everything these men learned going through doors, managing teams, and serving their communities translates directly into relationship-based sales, leadership under pressure, and building something that outlasts the badge. If you've ever told yourself you're not capable of anything outside law enforcement, this one is going to challenge that story hard.In this episode:• Moral injury is NOT PTSD: it is the cumulative damage caused when you perpetrate something, fail to prevent something, or are ordered to act against your moral compass, and it compounds silently over an entire career without a single critical incident ever being the trigger• Culturally competent clinicians are practitioners who have lived the lifestyle or are married to someone who has: Dewayne walked out of his first VA appointment when the counselor opened with pronouns, then found a female therapist married to a 20-year infantry veteran and experienced a completely different outcome• The American Warrior Association's R3 program is building proactive resilience frameworks inside agencies, including regional resiliency coordinators, vetted clinician networks, and fully funded Warriors Refuge retreats where flights and lodging are free to eligible officers and spouses• Brad Ortiz's Sound Off program at Silencer Shop donates $2 per purchase to a vetted nonprofit fund, and every individual officer purchase routes through silencershop.com where buyers work directly with Brad and Chris• The inflection point analogy: moral injury deposits are made silently every shift, just like compounding interest, and what looks like a single critical incident blowing up is actually the hockey-stick moment on a years-long accumulation nobody was tracking• Social media pile-ons after a justified OIS caused Brad more lasting moral injury than the shooting itself, a blind spot no clinician or debrief ever addressed until he found the right cultural fit in counseling• Baseline blood panels and brain mapping done at the academy level would give officers a before-and-after reference for physiological change across a career: the greater endocrine system and brain health are being completely ignored while TRT gets all the attention• The law enforcement skill set transfers directly into business leadership: problem-solving under pressure, decision-making without perfect information, ownership of mistakes, and relationship-first sales are all things operators have already built on the job, they just don't recognize it as a resumeChapters:0:39 Iron Sights mission statement and welcome2:02 Introducing Brad Ortiz and Dewayne Manson2:49 Brad's background: The Uncommon Line podcast9:40 American Warrior Association programs overview11:43 Resilience training inside TTPOA: the first of its kind12:08 What moral injury actually is and why it's not PTSD14:20 Culturally competent clinicians: Dewayne's personal story31:52 Compounding critical incidents over a career38:37 Social media pile-ons as a hidden source of moral injury44:55 Identity vs. calling: challenging how officers define themselves1:12:18 Relationship-first sales, legacy brands losing ground, and the business parallel1:21:46 How to access AWA, R3, and SoundOff resourcesMentioned:Brad Ortiz — Director of Sales for the Law Enforcement Division at Silencer Shop, former law enforcement officer and fugitive investigator, and former host of The Uncommon Line podcast, joined as a guest to discuss the SoundOff program and the AWA partnership.Dewayne Manson — Representative of the American Warrior Association, infantry veteran who served in Afghanistan and contracted there through 2020, shared his personal journey through moral injury and recovery as context for AWA's R3 program.Matt Smith — Co-host of the TTPOA podcast with 21 years on SWAT, co-hosted this live conference episode alongside Scott Howell and competed in the TTPOA shooting competition the day prior.Brandon Hernandez — Region 7 Director and Director of Training for TTPOA, co-host of the TTPOA podcast, referred to on air as the celebrity of the group when a passerby stopped to greet him.Anna Heil — Associated with the American Warrior Association, attended a TTPOA show in St. Louis and briefed other state organizations on the AWA partnership afterward.Chris — Works alongside Brad Ortiz at Silencer Shop on the SoundOff individual officer program, mentioned as a direct point of contact buyers will reach at soundsoftshop.com.Neil Noakes — Former Chief of Police of the Fort Worth Police Department, cited as an example of a career officer who, when asked how many critical incidents he had been on, simply said he had lost count.Simon Sinek — Author and speaker known for the concept of finding your 'why,' referenced when Brad Ortiz challenged every officer in the room to examine the deeper reason behind their career and identity.

The episode identifies a core structural shift in the managed services industry: the decoupling of service measurement from observable work due to the adoption of autonomous service desk technologies. This shift is driven by the introduction of automation platforms—such as Atera's Robin, Acronis AI Service Desk, and NinjaOne's endpoint automations—that eliminate or obscure traditional service tickets, shifting operational baselines and the metrics used for client billing and value demonstration. Evidence of this shift includes Atera guaranteeing that within 90 days, its Robin system will autonomously resolve half of Tier 1 and complex Tier 2 tickets, enforced via commercial contract terms. The company builds baselines by requiring six months of client ticket history before implementation. Supporting data from Channel EDE and AT&T show that automation can suppress visible ticket volume while inflating claims of efficiency and avoided incidents, independent of provider-side measurement. According to Dave Sobel, AT&T tracked autonomous incident handling since 2018, but most MSPs lack comparable historical data. Further developments reinforce this transition: NinjaOne integrates with ServiceNow to create incidents without human intervention, while Ingram Micro channel feedback observes partners aiming to increase business without staff growth. Broader labor market data and user sentiment surveys reveal that AI-backed automation does not show aggregate productivity gains (Stanford Economic Policy Institute) and is generally viewed with skepticism: Gallup and Apistevist data highlight declining confidence in corporate AI deployments and increased worker nostalgia for pre-automation workflows. The operational impact for MSPs centers on data ownership, measurement accountability, and renewal risk. As traditional records like tickets are eliminated or fragmented, providers who lack their own carefully preserved baselines may find themselves forced to rely on vendor-generated claims for demonstrating avoided work or cost savings. This creates exposure to contract risk, compromised pricing leverage, and governance complexity—especially if ticket-level detail, taxonomy, or supporting operational notes are lost in platform migrations or poorly configured retention policies. According to Dave Sobel, preparing by exporting comprehensive ticket histories, freezing operational taxonomies, and independently counting non-ticket sources of demand are now urgent requirements to maintain accountability and defensible value in future client negotiations. 00:00 The Ticket Is Disappearing  03:45 You Can't Invoice an Absence  07:20 The Client Already Stopped Believing 11:18 Why Do We Care?    Supported by: ScalePad

Quick update episode before I head out to walk the Camino! A few episodes ago, I told you about coming off my medication and sleep aids, the brain zaps, and the nights I barely slept.Here's where things landed: I'm fully off everything, sleeping better than I have in months, and I'm sharing exactly what worked.I'm also telling you about the day I broke my own rule, paid for it until 12:30am, and what I did the next morning — because that part is the whole lesson. If you've ever blown a boundary with alcohol and used it as a reason to give up on the whole thing, this one's for you. In this episode, you'll learn: The full update on coming off sleep aids and medication, including the last little crutch I didn't even realize I was using The two simple changes that made the biggest difference in my sleep What happened the day I broke my own caffeine boundary (and why it felt exactly like drinking) How to get back to a boundary after you break it, without the shame spiral or the "I'll start Monday" trap Why your body knows how to get back to baseline when you stop interfering Quick reminder: this is my personal experience with medication. Always talk to your doctor before starting or stopping anything.Also, I recommend you listen to 'How to Stop Talking Yourself Out of It When It Gets Uncomfortable' before you listen to this one, to get the full context! Listen to that here: https://www.angelamascenik.com/podcasts/stop-over-drinking-and-start-living/episodes/2149227702Now accepting applications for the next cohort of the Transformation Program, check that out here: https://www.angelamascenik.com ABOUT ANGELA: Angela Mascenik is a certified stop over-drinking coach for women and the host of the Stop Over-Drinking and Start Living podcast. She helps high-achieving women get to the root of why they drink and change their relationship with alcohol from the inside out — without white-knuckling it or relying on willpower. Angela is the creator of the 6-month Transformation Program, the Alive AF! membership, and The Magic House Retreat Center in Lisbon, Portugal.

Dr. Peter Dionisopoulos is a physical therapist and performance rehabilitation specialist focused on injury prevention, rehabilitation, and performance optimization. He emphasizes the value of finding the right exercise threshold where discomfort exists but pain doesn't, and the importance of progressive exposure to stress in a safe manner to build resiliency.  He shares that different body types have genetic predispositions toward certain movement patterns, the need for individualized approaches rather than one-size-fits-all solutions, and the significance of maintaining fitness levels even during recovery from injuries or surgery. Pete explains his approach at Dynamic Performance Rehab, which bridges basic rehabilitation with performance enhancement to help people achieve optimal function rather than just returning to normal. To connect with Dr. Peter Dionisopoulos visit dynamicprri.com or follow on Instagram @dynamicpr.ri   Visit ConfidenceThroughHealth.com to find discounts to some of our favorite products.Follow me via All In Health and Wellness on Facebook or Instagram.Find my books on Amazon: No More Sugar Coating: Finding Your Happiness in a Crowded World and Confidence Through Health: Live the Healthy Lifestyle God DesignedProduction credit: Social Media Cowboys

A polished report, monthly invoice or TEM dashboard doesn't necessarily give customers a baseline they can trust. In complex technology environments, contracts, invoices, inventories and ownership records often drift apart as services change, sites move and suppliers bill differently. When the baseline is wrong, strategy, sourcing decisions, savings targets, and budgets are built on shaky ground. In this episode of Staying Connected, Tony Mangino is joined by TC2's Frank Zagrodnik to discuss how enterprise customers can move from invoice fire drills to a defensible and actionable baseline by connecting contracts, invoices, inventory and ownership. If you would like to learn more about our experience in this space, please visit our IT Cost Management webpage.

Episode Summary: To secure the growing U.S. interests in space, it is imperative we have a capable and ready Space Force that can meet the challenges of the future. The Future Operating Environment 2040 and the Objective Force Baseline are the Space Force's assessment of what to expect and what forces and capabilities it will need to preserve space superiority. To learn more about these documents we had an in-depth discussion with two of the principal authors, Col. Paul Latour and Lt. Col. Sean Frederick. Credits: Host: Heather "Lucky" Penney, Director of Research, The Mitchell Institute for Aerospace Studies Producer: Shane Thin Executive Producer: Douglas Birkey Guest: Charles Galbreath, Director & Senior Resident Fellow for Space Studies, The Mitchell Institute Spacepower Advantage Center of Excellence (MI-SPACE) Guest: Jennifer "Boots" Reeves, Senior Resident Fellow for Space Studies, MI-SPACE Guest: Col. Paul LaTour Guest: Lt. Col. Sean Frederick Links: Subscribe to our YouTube Channel: https://bit.ly/3GbA5Of Website: https://mitchellaerospacepower.org/ Twitter: https://twitter.com/MitchellStudies Facebook: https://www.facebook.com/Mitchell.Institute.Aerospace LinkedIn: https://bit.ly/3nzBisb Instagram: https://www.instagram.com/mitchellstudies/ #MitchellStudies #AerospaceAdvantage #Space #Military #Future

See how a single Azure Database for PostgreSQL Flexible Server scales read-heavy workloads using read replicas and virtual endpoints — the same pattern OpenAI relies on to run ChatGPT on Postgres. Scott and Paula show a live read workload saturating a primary, then offload it to a replica with a single hostname change, and finish with replicas across Europe staying within milliseconds of the primary. It's read scale-out without sharding or re-architecting your app. Chapters 00:30 - Introduction 01:55 - Why read replicas: read/write asymmetry 03:45 - Architecture: primary, replicas, and virtual endpoints 05:30 - Demo: the cluster and a read-only replica 10:28 - Baseline: saturating the primary 13:53 - The flip: offloading reads to a replica 16:48 - Wrap up Recommended resources Read replicas in Azure Database for PostgreSQL Azure Database for PostgreSQL Demo script Connect Scott Hanselman | Twitter/X: @SHanselman Paula Berenguel | LinkedIn: paulaberenguel Azure Friday | Twitter/X: @AzureFriday Azure | Twitter/X: @Azure

See how a single Azure Database for PostgreSQL Flexible Server scales read-heavy workloads using read replicas and virtual endpoints — the same pattern OpenAI relies on to run ChatGPT on Postgres. Scott and Paula show a live read workload saturating a primary, then offload it to a replica with a single hostname change, and finish with replicas across Europe staying within milliseconds of the primary. It's read scale-out without sharding or re-architecting your app. Chapters 00:30 - Introduction 01:55 - Why read replicas: read/write asymmetry 03:45 - Architecture: primary, replicas, and virtual endpoints 05:30 - Demo: the cluster and a read-only replica 10:28 - Baseline: saturating the primary 13:53 - The flip: offloading reads to a replica 16:48 - Wrap up Recommended resources Read replicas in Azure Database for PostgreSQL Azure Database for PostgreSQL Demo script Connect Scott Hanselman | Twitter/X: @SHanselman Paula Berenguel | LinkedIn: paulaberenguel Azure Friday | Twitter/X: @AzureFriday Azure | Twitter/X: @Azure

Are your sleep trackers helping you sleep better? Or making you more anxious about your sleep? In this episode, we welcome back sleep expert Mollie Eastman, founder of Sleep Is A Skill, to explore how chasing the perfect sleep score can actually undermine restorative sleep, while sharing practical strategies for improving sleep through mindset shifts, Acceptance and Commitment Therapy (ACT), and smart use of wearable data. Mollie also discusses overcoming the "First Night Effect" while traveling, when to consider at-home sleep studies, and the foundational tests that matter most before investing in advanced biohacks. Finally, she shares her personal experience with psychedelic-assisted therapy at Beckley Retreats, revealing how addressing suppressed emotions and unresolved stress transformed both her sleep and overall well-being.Mollie Eastman is the creator of Sleep Is A Skill and the host of The Sleep Is A Skill Podcast. Sleep Is A Skill is a company that optimizes people's sleep through a unique blend of technology, accountability, and behavioral change. After navigating insomnia while traveling internationally, she created what she couldn't find - a place to go to learn the skill set of sleep. With a background in behavioral change from The Nonverbal Group, she became fascinated with chronobiology and its practical application to sleep and our overall experience of life. Knowing the difference between a life with sleep and without, she's now dedicated her life to sharing the forgotten skill set of sleep. In the spirit of that goal, she has created the #2 sleep podcast, where she has interviewed over 200 sleep experts, written a popular weekly sleep newsletter for over six years, partnered with luxury hotels & lifestyle brands, coached the world's top poker players, and has appeared on over 175 podcasts.SHOW NOTES:0:39 Welcome to the podcast!2:16 About Mollie Eastman3:19 Welcome her back to the show!4:48 Are wearables helping or harming?7:04 Placebo-Nocebo effect11:25 Getting relief from sleep data14:00 The ‘mindset' undermining sleep16:13 Acceptance & Commitment Theory (ACT) for Insomnia18:38 Using biohacks for “First Night Effect”25:05 Sleep disorders & at-home sleep studies32:14 Auditing sleep data36:13 Baseline tests to start with39:51 Her latest discovery on sleep support40:55 Beckley Retreats48:47 The benefits of psychedelics on her sleep54:24 The impact of suppressing emotions58:57 Where to find herRESOURCES:Website: Sleep Is a SkillSleep Obsessions' Monday Newsletter8-Week Wearable Group “Optimize Your Sleep” ProgramSleep Is A Skill PodcastLinkedInIG: @mollie.eastmanSleep Tests:SleepDoctor.comSleep Image RingHappy RingSleep Doctor Watch PAT - WatchPATOnera Sleep StudySupport this podcast at — https://redcircle.com/biohacker-babes-podcast/donationsAdvertising Inquiries: https://redcircle.com/brands

Today, I'm joined by Marco Suvilaakso, co-founder & co-CEO of Nucu.   Swapping wearables for "nearables," Nucu's low-touch ambient sleep monitoring platform gives guardians insight into kids' and teens' rest patterns.   In this episode, we discuss filling the market gap in pediatric sleep.   We also cover:   Building family-focused healthtech Benefits of real-time hypnogram data Avoiding obsession over metrics   Subscribe to the podcast → insider.fitt.co/podcast  Subscribe to our newsletter → insider.fitt.co/subscribe  Follow us on LinkedIn → linkedin.com/company/fittinsider    Nucu's Website: https://nucuhealth.com/  Nucu's Instagram: https://www.instagram.com/nucuhealth/  Nucu's LinkedIn: https://www.linkedin.com/company/nucu/    -   The Fitt Insider Podcast is brought to you by EGYM. Visit EGYM.com to learn more about its smart fitness ecosystem for fitness and health facilities.   Fitt Talent: https://talent.fitt.co/  Consulting: https://consulting.fitt.co/  Investments: https://capital.fitt.co/    Chapters: (00:00) Introduction (01:33) Marco background and Nucu overview (05:15) Kids/teens market gap  (06:30) Customer feedback and shift (09:54) Form factor redesign (11:20) Sleep consciousness building (15:03) Baseline and trend tracking (16:00) Room conditions correlation (18:31) Deep biometric insights (20:00) Training schedule impact (21:15) Lifespan data architecture (24:15) Ecosystem partnerships (26:30) Avoiding obsession over metrics (29:50) Real-time hypnogram data (33:40) Two target audiences (36:00) Research collaborations (37:07) Conclusion

In this episode, Hannah sits down with Taylor, a Special Education teacher at Little Light House, to explore a foundational approach to supporting Autism. Taylor walks parents through four key needs to check when dysregulation occurs: hydration, a full belly, rest, and health. She also unpacks the concept of sensory rest and why it looks different for neurodivergent children. If your child frequently struggles with dysregulation, this episode gives you a simple, powerful place to start. Every kid deserves the chance to just be a kid.At Little Light House, that belief drives everything we do. We provide tuition-free education and therapeutic services, rooted in Christ-centered care, for children with special needs and the families who love them.Learn More: https://www.littlelighthouse.orgLet's stay connected:Facebook: https://www.facebook.com/llhtulsaInstagram: https://www.instagram.com/llhtulsaBe part of the story:Give: https://www.littlelighthouse.org/give-helpJoin THECREW: https://www.littlelighthouse.org/the-crew

We were promised that more data would give us more control. Somewhere between the sleep score and the blood panel, many of us stopped asking how we actually feel. After stepping away from the demanding executive career that had shaped much of her identity, physician and healthcare leader Dr. Nasim Afsar was forced to confront a deeper question: Who are you when the role, schedule, and story that defined you are suddenly gone? In this conversation, Nasim and Marc explore identity beyond achievement, the danger of trying to fix every uncomfortable emotion, and the growing anxiety surrounding wearables, supplements, testing, peptides, and longevity protocols. They also examine how unified health data and artificial intelligence could create a more personalized and human healthcare system—provided that people remain the owners of their information and the centre of every decision. This is a conversation about becoming whole, trusting how you feel, and remembering that no technology can replace sleep, movement, purpose, community, and a life you genuinely want to live. Show Partners: Get your MENTAL FITNESS BLUEPRINT here! A special thanks to our mental fitness + sweat partner Sip Saunas Personal Socrates: Better Question, Better Life   Connect with Marc: https://konect.to/marcchampagne   Timestamps: 00:00 — The question that opens every interview: “Who are you?” 01:41 — Leaving the career that had become part of her identity 04:23 — What story are you telling yourself about who you are? 06:32 — Becoming a more whole human being 08:21 — What a Valentine's Day disappointment taught Nasim about pain 11:14 — Why fixing the problem can prevent us from understanding it 13:16 — Letting discomfort teach you instead of managing it away 15:38 — How medical training conditioned doctors to disconnect emotionally 18:36 — Feeling difficult emotions without becoming overwhelmed by them 20:03 — Why high-stress professions must teach people how to process trauma 22:17 — The connection between human context and intelligent health 23:47 — The patient labelled “non-compliant” whose real story was being missed 26:06 — Why clinical care represents only part of what shapes our health 26:44 — The three pillars of intelligent health 28:53 — Why modern health optimization has become overwhelming 30:12 — When wearable data disconnects you from how you actually feel 31:58 — The psychological cost of forcing yourself through “healthy” routines 33:34 — The anxiety and stress hiding inside the longevity movement 35:33 — A healthier, seasonal approach to wearables and tracking 37:56 — The simple fundamentals shared by people who live well 39:12 — The dangerous side of rushing into unverified health treatments 40:32 — Returning to the fundamentals before chasing the latest protocol 42:01 — Why fragmented health data prevents personalized care 45:15 — Where to begin when health information feels overwhelming 46:06 — Baseline testing, trusted practitioners, AI, and health-data privacy 47:38 — Why no peptide can replace the foundations of health 48:18 — The personalized health future that may be closer than we think 49:55 — Reclaiming autonomy without chasing every health trend * Special props

https://coachcollins.podia.com/funnel-down-defense https://teachhoops.com/ Today's episode is brought to you by TeachHoops.com, home to thousands of organized drill videos, defensive breakdowns, and course templates — everything in one click instead of scrubbing through random YouTube videos. In this episode, we install the LockLeft defensive framework: a simple, teachable system built on one ruthless idea — force every ball handler to his left hand, every possession, all game long. We cover the core rules: how on-ball defenders angle their stance to take away the right hand, how help defenders shade to shrink the left side of the floor, and how the whole scheme squeezes passing windows until offenses start throwing the ball to you. You'll learn the three most common breakdowns when teams first install it, how to rep it with simple shell drill constraints, and why this system is especially powerful at the high school and youth levels, where almost nobody can finish with their weak hand. If you've ever wanted a defensive identity your kids can master in two weeks and ride all season, this is the episode. Get the companion drill videos and full defensive install resources at https://teachhoops.com/. Learn more about your ad choices. Visit podcastchoices.com/adchoices

What does the data tell us about the real path to becoming an architect, and how is NCARB shifting its model to build a more adaptable, modular roadmap for future candidates?In this episode of Practice Disrupted, host Evelyn Lee breaks down the 15th edition of the NCARB by the Numbers report. She is joined by Jenny Kawecki, who leads data analytics and research at NCARB, and Gabriella Bermea, a licensed architect and NCARB volunteer who chaired the Experience Committee. Together, they explore the shifting dynamics of the architectural pipeline, unpacking why the median time to licensure has dropped to 12.3 years and what these statistics mean for the future of the profession.Jenny provides an insider's view of how NCARB uses this annual data report to track attrition points and actively improve its programs. She highlights major programmatic changes driven by historical data trends, such as retiring the rolling clock and revamping the Architectural Experience Program (AXP) reporting policy to give thousands of candidates their experience credits back. Gabriella adds an on-the-ground perspective, explaining how challenging past assumptions about rigid timelines helps eliminate systemic barriers and re-engages candidates who have stepped away from the path.The conversation also digs deeply into NCARB's massive "Pathways to Practice" initiative, a multi-year effort to pivot away from a single, rigid route toward a highly modular, competency-based framework. They address critical pipeline pinch points, including firm culture support during the AXP and the stark demographic pass-rate disparities within the Architect Registration Examination (ARE)."Instead of pushing everyone through the same path, how can that path meet the needs of the individual? How can it be customizable almost to the candidate so that it aligns with your background, your career path, your educational path, your social path. We want there to be a licensure path that exists for you." - Jenny KaweckiFinally, they look forward to how a more flexible, customizable licensure path can help mitigate the steep financial burdens of higher education and foster a significantly more diverse, representative body of architects.Guests:Jenny Kawecki is NCARB's Assistant Vice President of Data, Analytics, and Research, where she leads a team that analyzes industry trends and oversees publications like NCARB by the Numbers and Baseline on Belonging. Before joining NCARB in 2016, she wrote for Barnes & Noble and SparkNotes, and she holds a degree in Media, Culture, and the Arts from The King's College.Gabriella Bermea, AIA, is a Senior Associate and Architect at Perkins Eastman with more than eight years of experience in educational design, stakeholder engagement, and community-focused solutions. A fifth-generation Tejana, she is a passionate advocate for educational equity and community empowerment and currently serves as Chair of NCARB's Experience Committee.This episode is especially for you if:✅You want to know why the median time to licensure dropped and how policy changes like eliminating the rolling clock shifted the needle.✅You are curious about NCARB's "Pathways to Practice" initiative and how a modular, competency-based framework will offer diverse alternatives to the traditional degree path.✅ You want to explore the primary pinch points in the licensure pipeline, from firm culture barriers during the AXP to pass-rate disparities within the ARE.✅You want to learn how the revised AXP reporting requirements allow candidates to receive 100% credit for hours up to a year and 75% credit indefinitely.✅You are interested in how scholarship programs through NOMA chapters and state boards are helping candidates facing financial and institutional adversities stay on the path.What have you done to take action lately? Share your reflections with us on social and join the conversation.

Setting a Baseline for Cam Ward w/ Lucas Panzica | Football & Other F WordsSee omnystudio.com/listener for privacy information.

Welcome to the grand finale and weekly review of our master series. In this closing session, we explore the advanced neuro-somatic practice of permanent nervous system baseline integration, focusing our ultimate awareness on the 7th Chakra (The Crown Center). Discover the cutting-edge science of how a seven-day commitment to somatic tracking leverages neuroplasticity to structurally rewrite your default stress responses. By weaving the foundational elements of your week into our crowning, integrated protocol, you will learn to permanently anchor your nervous system on an inner throne of sovereign stillness. Turn off the noise of the outside world, claim your absolute crown, and drift effortlessly into a deep, restorative night of sleep.