Podcast
Latent Space: The AI Engineer Podcast
Hosted by Latent.Space · EN
The podcast by and for AI Engineers! In 2025, over 10 million readers and listeners came to Latent Space to hear about news, papers and interviews in Software 3.0.
We cover Foundation Models changing every domain in Code Generation, Multimodality, AI Agents, GPU Infra and more, directly from the founders, builders, and thinkers involved in pushing the cutting edge. Striving to give you both the definitive take on the Current Thing down to the first introduction to the tech you'll be using in the next 3 months! We break news and exclusive interviews from OpenAI, Anthropic, Gemini, Meta (Soumith Chintala), Sierra (Bret Taylor), tiny (George Hotz), Databricks/MosaicML (Jon Frankle), Modular (Chris Lattner), Answer.ai (Jeremy Howard), et al.
Full show notes always on https://latent.space
www.latent.space
www.latent.space
110episodes
Episodes
Newest firstAll episodes
Reality: The Final Eval — Lukas Petersson and Axel Backlund of Andon Labs
5d ago01:15:39Summary readyThe new AIEWF website is live! Get your tickets booked ASAP as they -will- sell out. Take the AI Engineering Survey and get >$2k in credits and free AIE WF tickets!Most industry benchmarks compress intelligence and reasoning ability into scores.SWE-Bench Pro, MMLU, Humanity’s Last Exam, etc. These metrics are useful, but don’t always represent the full extent of how a model performs in the real world. Some of the most interesting evals today look less like exams and more like operating businesses in the real world. One of which is Vending Bench.In Anthropic’s Mythos Preview System Card, Andon was the only third party eval to get their own section, observing increasingly concerning aggressive behavior:You don’t know what a model is capable of doing in the real world unless you actually give it inventory, a wallet, tools, customers, competitors, humans, & some time. More often than not, it’ll surprise you how much a model is capable of and in doing so, also reveal unexpected behavior: deception, context collapse, emergent coordination, & bizarre negotiation behavior.While an inflection point in personal agents came post-OpenClaw after full file access with bypass permissions became the norm, it is yet to come for agents in the real-world. However Andon Market, an actual in person store fully run and managed by AI, is paving the way for what is possible.Full Video PodFrom Claude trying to call the FBI over a $2/day vending machine charge to AI agents forming price cartels, hiring human employees, running physical stores, and writing existential robot musicals, Andon Labs is stress-testing what happens when frontier models stop being chatbots and start acting in the real world. In this episode, Andon Labs cofounders Lukas Petersson and Axel Backlund join swyx and Vibhu to unpack the strange, funny, and genuinely concerning edge cases that emerge when agents run businesses over long horizons.We go deep on Vending-Bench, Project Vend, Vending-Bench Arena, Bengt, Butter-Bench, Luna, and Andon’s broader mission of building realistic real-world evals for autonomous AI systems. Lukas and Axel explain why dollar-denominated evals reveal things traditional benchmarks miss, how Claude ended up reporting its vending machine fees as cybercrime, why long context windows can drive agents into meltdown loops, what happens when agents compete with each other, and why the future of AI safety may depend on testing models in messy physical environments instead of clean benchmark sandboxes.We discuss:* Why Andon Labs started with dangerous capability evals and long-running agents* Vending-Bench and why running a vending machine is a deceptively hard AI benchmark* Why money-based evals avoid the saturation problem of traditional benchmarks* How Claude tried to call the FBI over a $2/day fee* Why long-horizon agents can spiral into existential and legalistic breakdowns* Project Vend: putting an AI-run vending machine inside Anthropic* Why real humans are “out of distribution” for simulated agents* Claudius, Seymour Cash, and the chaos of AI CEOs* How a human briefly became CEO of Claudius through a manipulated election* Why multi-agent systems can converge back into “helpful assistant” behavior* Bengt, Andon’s internal office agent with email, spending, terminal, phone, camera, and internet access* How Bengt traded Amazon purchases for face-recognition training data* Claude’s aggressive behavior, lies, refund avoidance, and price-cartel behavior in Arena* Why eval awareness may become the AI version of “are we living in a simulation?”* Blueprint Bench, spatial intelligence, and why models still misunderstand physical rooms* Butter-Bench and testing LLMs as robot orchestrators* Luna, the AI-run physical store with a three-year lease and human employees* The new Andon cafe in Sweden and why real-world geography matters for agent evals* Rotten tomatoes, perishable goods, and the hidden difficulty of running a physical businessLukas Petersson* LinkedIn: https://www.linkedin.com/in/lukas-petersson-181a83172/* X: https://x.com/lukaspetAxel Backlund* LinkedIn: https://www.linkedin.com/in/axelbacklund* X: https://x.com/axelbacklundAndon Labs* Website: https://andonlabs.com* Vending-Bench: https://andonlabs.com/evals/vending-bench* Andon Vending: https://andonlabs.com/vendingTimestamps00:00:00 Introduction00:01:00 Andon Labs and the Origins of Vending-Bench00:05:21 Why Money-Based Evals Matter00:09:51 Agent Harnesses and Self-Modifying Systems00:13:36 Claude Calls the FBI00:16:33 Project Vend: Claude Runs a Real Vending Machine00:21:44 Seymour Cash, AI CEOs, and Election Chaos00:27:16 Multi-Agent Coordination and Slack Observability00:30:18 When Will Agents Run Real Businesses?00:34:56 Bengt: Andon’s Internal Office Agent00:40:06 Real-World AI Safety and Long-Horizon Traces00:44:28 Lying, Refunds, and Price Cartels in Arena00:52:42 Eval Awareness and Simulation Behavior00:56:06 Blueprint Bench, Butter-Bench, and Robotics01:04:37 Luna: The AI-Run Physical Store01:09:29 The Sweden Cafe and Real-World Expansion01:13:16 What Comes Next for Andon LabsTranscriptIntroduction: Andon Labs, Long-Running Agents, and Real-World EvalsSwyx [00:00:00]: Welcome to Lukas and Axel from Andon Labs, and I’m joined by my, favorite guest host. Anything security, safety, alignments, Vibhu., welcome.Lukas [00:00:15]: Thank you for having us.Axel [00:00:16]: Thank you.Swyx [00:00:17]: Let’s match names to voices., maybe you wanna take turns introducing yourselves.Lukas [00:00:21]: I’m Lukas.Axel [00:00:22]: And I’m Axel.Swyx [00:00:24]: Let’s introduce Andon Labs a bit. How did you guys come together?, you have different backgrounds, but you’re both Swedish., was that, a big part of it?Lukas [00:00:33]: So when I went to high school, there was this really cool guy who had a superpower. He could code. So he made like the or like the app for the, for the school and stuff, and he was super cool, and I wanted to be like him, and that was that guy.Axel [00:00:47]: I don’t know about this.Swyx [00:00:49]: But you went to different universities, right?Lukas [00:00:51]: But same high school.Swyx [00:00:52]: I see.Lukas [00:00:52]: So we always said, “Oh, once we graduate university, then we should start a company,” and that’s what we did.Swyx [00:00:58]: Wow, there you go. And about a year ago, you kinda burst onto the scene with Vending Bench, but, was there a thing before that was, kind of like the inception?From Dangerous Capability Evals to Vending BenchAxel [00:01:07]: So we did work, yeah, with, Anthropic was one of our, early customers in doing, evals. So we did, dangerous capability evals., nothing we published openly. But then we started thinking about do...
Summary
🔬Scaling Past Informal AI - Carina Hong, Axiom Math
6d ago01:33:04Summary readyIn 2025, seven-month-old startup Axiom solved all 12 of the problems Putnam exam (scoring 8/12 in the time limit) a prestigious undergraduate math exam. The 12/12 score is better than the top undergraduates (110/120) and the closest AI system that reported a result (DeepSeek 103/120), although it is unclear what the people and other systems would have scored with more time. Nonetheless, the Putnam exam is legendary for its difficulty, with the median score typically being 0 or 1 points. Taken by itself, this seems like a minor feather in the cap of AI; one of a long series of accomplishments by AI systems in elite competitions with humans, starting with Deep Blue beating Kasparov.Fast forward to mid-2026, and Claude Code is eating the world. In 2024 Anthropic’s bet on code and enterprise looked like a more pragmatic niche play vs. OpenAI’s better models and massive consume scale. Today, Amodei’s all in bet on acceleration via code (images and video be damned) seems prescient.Despite Anthropic’s growing momentum, however, Axiom CEO Carina Hong sees coding ability as a necessary but not sufficient milestone on the path to AGI. Code arguably pushes the jagged frontier to the point of super intelligence in some domains outside of coding, but there are surprising gaps (link) that Carina believes will bottleneck AI progress. (Stats on math benchmarks).The informal bottleneck“Verified AI” sounds like eating broccoli (footnote: I actually love broccoli, but then again, I also believe strongly in Test Driven Development, so ¯\(ツ)/¯ ) and paying taxes, but to Axiom it means something very different. “Verification to me is about scaling brilliance, compounding brilliance,” Carina told us.It actually took a while for me to understand what she means by this. It sounded like marketing-speak to me, until it clicked. Carina emphasizes an story about legendary mathematician Srinivasa Ramanujan to illustrate the point. When G.H. Hardy finally persuaded Ramanujan to formally prove theorems instead of relying on his (formidable) intuition, it reportedly improved his own capabilities. This is presumably because formally proving things forced Ramanujan to articulate the details in a way that open up new lines of thinking, etc. This is one part of “compounding.”But formally proving things also allowed others to benefit from his intuition: the proofs are way of communicating an intuition and persuading others that the intuition is correct. This is scaling (more people use the result) and compounding (people can learn from and build on his work).This is the analogy that Carina wants us to focus on.Verified GenerationThere are two ways that Verified AI shows up: in training and in inference.But a quick detour: to a first approximation, “Formal Verification” means using type checkers (like for TypeScript, C++ or Rust, but more capable) to verify mathematical proofs that are meticulously specified using a language like Lean (footnote: Formal verification also includes model checking (TLA+, SPIN), SMT-based tools (Dafny, F*, Why3), and refinement-type systems (Liquid Haskell) — many of which don’t look much like “type checking a proof” from the user’s perspective even when there’s a similar logical core underneath. It also gets applied to software and hardware correctness, not only pure mathematics.). It takes a lot of work to translate an “informal” proof (albeit one that most people would not remotely call “informal”) in to a Lean proof (footnote: This is an understatement. Most theorems remain informal because formalization is so hard to do. There has been a great deal of effort to formalize the most important proofs, with mixed results)You can imagine how this would be (very) useful during Reinforcement Learning: instead of relying on best guesses based on statistics (GRPO, RLHF, etc.), you can just verify the proof is correct using a Lean verifier. This is obviously a much stronger reward signal, akin to compiling code and testing it (which is what is typically done with RL on coding).The catch: LLM are not (currently) very good at proving things with Lean.Enter Axiom: While they have not officially reported benchmark numbers besides the 12/12 Putnam result, Carina reports that they have achieved a very impressive 99% (187/189) ProofGen on the Verina benchmark. This benchmark is to generate code and proof of correctness for a series of problems. For context, OpenAI o3 (the last known OpenAI run) achieved 4.9% on this benchmark.Based on the sparse benchmarking, it’s hard to say what the frontier labs are currently doing, but Carina suggests that they still are not training to generate Lean proofs directly, rather relying on informal proofs.Time will tell if the frontier labs’ current approaches will close this gap.Scaling and compoundingCarina’s Ramanujan analogy is pretty direct. Better proofs → better Lean generation → better RL. A stronger signal means higher sample efficiency and higher maximum performance. Great!Scaling is pretty clear too: once I have proved something in Lean, the quality of the output is basically (footnote: one might argue that its a bit lower because the proof is in distribution for the LLM) as high as if it came from a human, so my high quality training set has grown in a way that an informal rollout corpus cannot. I can trust my Lean proofs.Compounding is also clear: now all of future inference and training can build upon those proofs.On the other hand, a model trained only using statistical signals like GRPO during RL lacks the sample efficiency, maximum performance and compounding corpus that a system that uses formal verification benefits from.All roads lead to verificationBroccoli and taxes notwithstanding, “verification” has shown up in a lot of conversations recently. In the in physical system control:“I think [verifiability] is probably the hardest problem right now, because the as the models get better, it can be harder and harder to find the faults on the system. And so the problem of doing proper eval to find those faults, that problem also keeps getting harder as the models get better.” -In theoretical physics:“…now that we’re in this regime where you can just get ChatGPT to tackle thousands of questions at the same time, it will return proofs for a significant fraction of them. Now actually the onus is back on the humans to verify all the outputs. And so, yeah, as that becomes a bottleneck, I think formalizing math and automating verification will become more valuable.” -Verification is, in fact, the key differences between AI for science and AI for computation: in science you to have to actually test (verify) your hypothesis by performing physical experiments. Lab in the loop systems like Radical AI and Lila build around exactly this premise (we have recorded episodes with both of these teams and will release them soon!)And yes, formally verifying critical systems such as flight control, nuclear power plants and pacemakers is a growing focus as the software and hardware that run them becomes more complex.Carina believes so strongly that AGI requires verified generation that she makes the unqualified claim that “We do not believe there is any other possible future.”Expensive to produce, cheap to verifyLean proofs are hard generate, but they can be easily shown to be correct or incorrect. But how do you know that the proof you created maps correctly to the problem you care about? As Carina puts it: “Anything that can be specified can be proven. Humans are bad at specifying everything we want.”Are we now in the specification business? Check out the episode to hear Carina’s take, as well as:* Why hardware verification is a killer app* Details on the AXLE open API and recently released Discovery toolkit* The Erdos debacle* The OpenAI GPT-f diaspora This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit www.latent.space/subscribe
Summary
⚡️Satya Nadella: No Priors x Latent Space Crossover Special at Microsoft Build
6d ago00:38:58Summary readyWe’ve informally heard that Satya is a listener to LS for a couple years now, but it was still absolutely surreal to meet him and do a live pod at Build, together with our friends at No Priors, the leading VC AI Podcast that we also greatly admire!We covered the MAI model technical takeaways on yesterday’s AINews, so I will focus our recap of Satya’s main messages around three elements:* Satya’s adaptation of the Bill Gates Line for positioning Microsoft as the Frontier Intelligence Platform — customers must gain much more value from the Microsoft ecosystem than Microsoft itself, by building on multi-model harnesses like OpenClaw and Scout, drawing on the full enterprise context exposed by context layers like Work IQ (heavily dogfooded by his C-suite), and building up private evals and traces as a new form of Token IP* AI ROI: On one hand, enterprises are having difficult conversations around Tokenmaxxing and Layoffs, and on the other hand, there are serious re-evaluations of the End of SaaS since the Build vs Buy equation has changed so much. Our previous SemiAnalysis guest had… interesting comments on Microsoft’s position on this as the ur-SaaS titan, and Satya had great answers* Making the Impossible Possible: Kevin Scott’s inspiring framing around what the most ambitious version of applying AI and technology at large to business and social problems, like education and social impact.Enjoy!Full VideoTranscriptVoiceover: Welcome swyx, Sarah Guo, Elad Gil,, and Chairman and Chief Executive Officer of Microsoft, Satya NadellaSarah Guo: Welcome to a crossover episode of No Priors and Lane Space with Satya Nadella. Um, congratulations on an amazing build. No, thank you so much, and it’s great to be with both of you. I listen to both of you or b- both the podcasts all the time. It’s great to be on it.Thank you so much. [00:01:00] So you’re just talking about, um, these amazing, uh, announcements from across the Microsoft estate all morning for, I think, three hours. What is the, uh, what’s the most important reflection or takeaway you have?AI as an Ecosystem PlatformSarah Guo: I, I’d say there are, uh, perhaps the, the biggest one for me is let’s sort of conceptualize this more as an ecosystem play as opposed to a single model or even a single platform, right?Satya Nadella: I mean, you know, whatever I... At least for me, having grown up at Microsoft, having seen, whatever, four major platform shifts, uh, I sort of fall into that, um, uh, camp where a platform is defined by fundamentally its ability to create more value about the platform versus what’s captured in the platform. And so if you, you view what’s happening right now, I think this morning’s keynote was how can any company, whether it’s an AI native company or a traditional enterprise company, participate as a first-class participant where they can point to AI they created, [00:02:00] right?It’s not that they don’t use other people’s AI. Of course they will. But to me, what’s the path? What’s the recipe? How do I do it? What does a stack look like? What does the tooling look like? What is valuable? How do you do that? That’s it. That’s sort of our job to do. Yeah. Ecosystem strategy is, uh, very complicated, right?Sarah Guo: Because you end up building certain components, partnering for certain components, supporting them. You just announced this big suite of models. Like, tell us a little bit about the, uh, training strategy for Microsoft now. Yeah.MAI Models & Training StrategySarah Guo: So, so the thing that we wanted to do with the MAI models was to build, and as Mustafa talked about, first of all, a great lineage, right?Satya Nadella: Starting with pre-training, uh, with very good data quality, uh, doing all the ablations, making sure because in, in some sense it’s becoming even harder to build a clean lineage model just because there’s so much stuff out there, uh, that you truly need to ablate out to be able to have a fantastic [00:03:00] pre-trained model.In fact, that’s one of the challenges of a lot of the open weight models is they look great on one benchmark or two, but they’re not great on practice. So that’s why, in fact, even in the RFDEs are, they, they are pretty gone really excited about these MAI models because how the heck can a small five B model hill climb?Uh, and it goes back a little bit to what I think is ultimately the key thing to do, which is try to pursue finding that cognitive core. Uh, so to me, starting with a clean lineage- Then creating that ability for companies to be able to use this, right? Not just as a generalist, but to create their own specialist by building this hill climbing scaffold around it, right?So it’s not just the model, but you have a hill climb scaffold around it, then you will start building your RLE. You will start collecting the traces. Most importantly, you’ll have private evals because we know all the evals out there are good, interesting, [00:04:00] but they’re not really that critical- They’re work, yeahSwyx: at this point because they all can be maxed. And so the point is each company will have its own private eval. And so that end-to-end platform story around our models is sort of, uh, what I think is interesting. And then the one other thing, Sarah, since you brought that up, is I do feel there’s a new frontier.Satya Nadella: Like people talk about the frontier and are you operating at the frontier. Um, interestingly enough, if you add a little temporality to it, you can use, let’s say, in, in, in fact, the, the Lando Lakes demo we showed was pretty cool. We used, whatever, GPT-55, right? Then you collected a bunch of traces, and then you took a 5B reasoning model and achieved higher.Sarah Guo: Uh, so that is another aspect of what it means to appear... uh, you know, operate at the frontier Yeah. I, I think, uh, I first of all have to congratulate you on basically building a frontier neo lab inside of Microsoft in two years. Um, I’m wondering, you know, you have all this AI strategy that you’re rolling out.Lessons from Two Years of AI DevelopmentSwyx: I’m wondering, what do you know now that you wish you would tell yourself two years ago where- or two or [00:05:00] three years ago? Three years for the Jensen partnership, two years for, uh, MEI. Yeah, I mean, I think the, the thing when, that I reflect quite a bit, right, which is sort of obviously I got into all this when I got excited by the, the scaling laws paper and, you know, when, you know, even the OpenAI partnership came about when those folks said, “Hey, we’re gonna really throw a lot of computer transformers.”Satya Nadella: Uh, and they’ve helped. I- the thing that I always look back and say, “Wow, these things, uh, do have capability that they’re climbing up.” W- I mean, this, you know, this crude way of saying it is intelligence is log of compute kind of works. Now what I think we underestimated perhaps is the real-world complexity of deploying these so that they actually deliver the value in the real world, right?So the outcomes as measured by any benchmark is interestingly important, but the true eval is when people out there are able to do unique things that they only can value, and it’s very [00:06:00] measurable, right? That I wish we had sort of even, like, had more in our consciousness, right? Which is as an industry.Sarah Guo: Because right now I think when people say, “Wow, I don’t want a token max,” it’s an artifact of us not having thought ourselves as an industry that we are using tokens to create value every step of the way. So I think that’s kind of what I wish we had gotten there, but I’m glad we are here.Real-World Value & Use CasesSarah Guo: What are some of the use cases that you’ve seen that have created the most value for your customers?Because I know that people talk a lot about code, and I think it’s pretty clear that that’s something that’s having very large scale impact. Are there other areas that you find in common that your customers are really benefiting from? Yeah. I think, yeah, to your point, obviously coding is now got... But it’s interesting, by the way, Elijah, to even talk about the coding, right?Satya Nadella: Which is coding has worked so well that we now have to rebuild the IDE, right? I mean, it’s kind of nuts to see what we sh- launched is like, oh my God, I have these hundred agent sessions. I... The cognitive load it transfers back to me as a human is so [00:07:00] excessive that now I need a new UI. Uh, oh, by the way, I, like the, the chat as the only artifact was also impossible, so that’s why we need a canvas.So it’s kind of interesting for all the things about where is software needed or where is UI needed, uh, you kind of need that even for code, right? In a fully agentic world. But that said, one of the things that we are starting to see, we started seeing with co-work, but even some of the work we, we showed with auto com- uh, um, autopilot Right on what you see with claws is a good one because if you sort of think about a lot of human capital is doing the glue work, rig...
Summary
GitHub's plan for Agents — Kyle Daigle, GitHub
1w ago01:23:27Summary readyI’m excited to work with Microsoft once again as the presenting sponsors of the AI Engineer World’s Fair! We’ll streaming live from MS Build today for a special crossover pod with our friends at No Priors and the one and only Satya Nadella. However we did not hold back with this interview - we asked all the burning questions about uptime and Copilot that we know you have in your minds. Lets go!For almost two decades, GitHub has been the home of software, where both open source and closed flow, through commits, pull requests, reviews, actions, etc.This ecosystem flourished as open-source maintainers and contributors would continue shipping code for the benefit of the community. However as coding agents began to ship mass quantities of code - growing 1400% in 2026, it marked a new era that was both extremely exciting and challenging for GitHub.While these agents help more people ship more projects, they also significantly increase the floor of how much code is shipped, how often it is shipped, how many people commit code, and basically orders of magnitude multiples in every dimension of GitHub infrastructure:Now GitHub inevitably experiences more pressure on their infrastructure which was originally designed around human developers moving at human speed. This has resulted in a very publicly notable uptime story:So it begs the question of whether current systems around code can absorb what AI produces. Can CI/CD keep up when every idea becomes a build? Can open source maintainers survive floods of AI-generated slop contributions? Can GitHub preserve the human social contract of software while becoming the operating layer for agents?Which brings us to the perfect person to answer these questions: GitHub COO Kyle Daigle. In this episode, he joins swyx to unpack what happens when AI doesn’t just autocomplete code, but starts changing how companies operate, how open source works, how pull requests get reviewed, and how GitHub itself has to scale. We go deep on GitHub’s internal AI workflows: micro-skills, WorkIQ, MCP, Slack, Teams, email, Copilot workflows, the new Copilot desktop app, CLI, cloud agents, and how Kyle uses agents to look backwards across company context before deciding what to do next. Kyle also reflects on GitHub’s history building webhooks, APIs, Actions, npm, Dependabot, and Semmle, why the AI era is breaking GitHub in new ways, how Actions became a general-purpose compute layer, and what Copilot becomes after code completion.Full Video PodWe discuss:* Kyle’s expanded role across GitHub* How AI got Kyle coding again after years in leadership* Why GitHub rolls out AI through existing workflows instead of forcing new tools* WorkIQ, MCP, Slack, Teams, email, and GitHub as company context* Why massive “mega-skills” are giving way to small, atomic micro-skills* How AI changes summarization, communications, marketing, and analyst work* Why former developers in leadership may have a unique advantage in the AI era* Kyle’s “15 agents on Saturday” workflow* How Kyle built an AI-generated executive presentation for CRO/CFO teams* Why AI changes the chief of staff role without removing the human work* GitHub Actions, webhooks, arbitrary code execution, and secure agent compute* The npm acquisition, supply-chain security, 2FA, and token invalidation* Slop forks, vendoring, and whether AI agents change dependency management* What pull requests become when most PRs come from agents* Prompt requests, vouching, AI review, and trust in open source* What counts as a “developer” when AI lowers the barrier to building* GitHub Spark, low-code, and why GitHub refuses to hide the code* 14x commit growth, Actions load, databases, monorepos, and availability* Copilot’s evolution from completion to CLI, desktop app, cloud agents, and SDK* Context, memory, rules, and making GitHub “act like Kyle wants it to act”* Ambient AI, OpenClaw, enterprise security, and the new operating system for agents* What swyx should ask Satya Nadella about Microsoft’s AI futureKyle Daigle* LinkedIn: https://www.linkedin.com/in/kyledaigle* X: https://x.com/kdaigleTimestamps00:00:00 Introduction00:03:36 Why AI Got Kyle Coding Again00:07:04 Running GitHub with AI: WorkIQ, MCP, Slack, Teams, and Skills00:15:39 The Golden Age for Former Developers in Leadership00:17:31 15 Agents on Saturday and AI-Generated Executive Work00:20:20 How AI Changes the Chief of Staff Role00:21:45 GitHub’s History: Actions, npm, Webhooks, and Open Source00:28:45 Slop Forks, Vendoring, and AI Dependency Management00:33:57 Pull Requests, Prompt Requests, and Trust in Agent-Generated Code00:41:21 GitHub Stars, 200M+ Developers, and the New AI Builder Wave00:45:15 GitHub Spark, Low-Code, and Why GitHub Still Shows the Code00:47:38 GitHub’s Hardest Era: 14x Growth, Reliability, and Scale00:59:21 Actions as the Compute Layer for CI/CD and Automation01:02:04 The State and Future of GitHub Copilot01:08:24 Ambient AI, Background Agents, and the Future of the SDLC01:13:09 OpenClaw, Enterprise Security, and the New OS for Agents01:18:03 Build Announcements, WorkIQ, FoundryIQ, and Microsoft Context01:21:41 What Should swyx Ask Satya?TranscriptIntroduction: Kyle Daigle’s Expanded Role at GitHub and MicrosoftSwyx [00:00:00]: We’re here with Kyle Daigle, COO of GitHub. Welcome.Kyle [00:00:07]: Hey, thanks for having me.Swyx [00:00:08]: You’re not just CEO of GitHub. People know you as that. You have a new role.Kyle [00:00:11]: So I have an expanded role now. I’ve been working at GitHub for thirteen years and doing all things developer. Joined as a developer myself. And now, I’m also responsible as the CMO of Developer for Microsoft. And so all the kind of learnings and passion for developers and how we work with them and how we communicate and how we bring our products to market, we’re also bringing that expertise to the broader Microsoft ecosystem and helping every developer that uses a Microsoft product or would like to have a sort of similar experience that they’ve had with GitHub over the years. So it’s a different role in some ways, but it’s also just building on the experience that I’ve had at GitHub of just sort of tell the truth, be authentic, show people how to use it and then let the products speak for themselves. Now just doing that with, all of Microsoft.Swyx [00:01:09]: We’ll be releasing this in conjunction with Build. You got lots of stuff planned, and we can sort of touch on that whenever it’s appropriate. I think one of the interesting things is I rarely meet a COO who’s also a CMO. I think you’re a very outward facing and you’re very confident publicly. That’s rare. Do you actually view yourself as COO? What’s What is your thing?From GitHub Developer to COO/CMO: Building the Platform and Operating GitHubKyle [00:01:33]: I think for me, it’s been funny. The titles have always been, a— have always felt a little strange to me. I joined GitHub as a developer? I wrote so much of theSwyx [00:01:46]: Let’s bring that up. You wrote the back ends?Kyle [00:01:48]: I was going through, I was going through, some old photos, when folks were talking about how things were being built or how there was a build GitHub. I built, webhooks and worked with teams building the API, built the platform layer. Anything that integrated with GitHub, up until really twenty eighteen, I built or ran the engineering teams. And that’s kind of where my the beginning of my passion always was helping people build things, deliver them to, their customers. And so being a developer, building for developers was always super unique. In a— I think as my role expanded, it became my ability to talk to not just developers, but also enterprise customers or business leaders and have this translation layer. And then through all those years, GitHub has always operated pretty uniquely. Post-pandemic, working remotely was not as novel as it was when GitHub started in two thousand and eight. But all that expertise of running remote teams, doing it well, became this sort of bigger role, ultimately turning into the COO role of how do we operate GitHub in the way that G...
Summary
Why Video Agent models are next — Ethan He, xAI Grok Imagine
1w ago01:43:26Summary readyWe’re announcing AIEWF speakers this week! Take the AI Engineering Survey!Today’s guest Ethan first joined us for the LS Paper Club as the lead on NVIDIA Cosmos World Model, but then joined xAI and built Grok Imagine in 3 months:He comes back on Latent Space with some nuclear hot takes: that Video Models primarily get their intelligence from LLMs, not from training on video data, and that the next frontier for truly interactive, realtime, long-horizon world models is to work on LLMs (perhaps Interaction Models as well…)Put it this way: In the near term, the next Sora won’t be a better video model, but a video agent.Generative Media may more closely follow the evolution of AI coding which went from focusing on one-shot output performance and cost, to multiturn reasoning and planning models for agents and systems that can plan, edit, test, debug, and submit PRs.At a certain point, coding models got so good that the only significant next step to improve performance was handling the orchestration of these models.Now as the performance of video models increases significantly across realism, consistency, & prompt adherence while becoming more cost efficient, the next evolution of video generation may also be systems that can plan, generate, edit, critique, and iterate across an entire creative task. In this episode, Ethan joins swyx and Vibhu to unpack what it actually takes to build frontier image and video systems: data, VAEs, diffusion transformers, audio-video alignment, inference speedups, and the hidden cost of storing and moving massive video datasets. From building NVIDIA’s Cosmos world model to joining xAI as Grok Imagine was being built from zero to one, Ethan He has been at the center of some of the most important work in video generation, multimodal models, and real-time world models.We go deep on Grok Imagine, how a small xAI team shipped its first multimodal video model in three months, why iteration speed matters more than almost anything in model development, and why many of the biggest gains come from fixing tiny bugs in data and training pipelines. Flipbook: The future of VideomaxxingVideo agents are almost a sure bet to be the trend in the coming year. We end with a glance at what’s beyond video agents:Flipbook caused a minor sensation this year when it was released, but most treat it as a fun demo. Ethan takes it very seriously — with the speed and cost of inference coming down every year, the future of custom video JIT UI is closer than you think. We talked about why videogen models may become the front end of AI, how generative UI could replace traditional HTML/CSS, why world models need to be real-time, interactive, and long-horizon, and why the future of video generation may depend more on language models and agents than on diffusion alone.We discuss:* Why fast iteration mattered more than meetings* Why small training bugs can drive huge model quality gains* Why coding models may make compute the bottleneck again* How image and video models are trained with synthetic captions* The role of VAEs and latent space in frontier video models* Why image models are the foundation for video models* The tradeoff between temporal compression and real-time interactivity* Flipbook, Neural OS, and the future of generative UI* Why future interfaces may go from user intent to pixels* The hidden cost of training video models: storage, egress, and GPU hours* How step distillation and consistency models (like OpenAI sCM) makes video inference orders of magnitude faster* Grok Imagine 0.9 and large-scale audio-video generation* Why audio-video alignment is harder than text-video alignment* Ethan’s definition of world models* Reference-to-video, video extension, and long-context video generation* Why xAI’s research communication undersells Grok Imagine* How xAI culture shaped the speed of development* AI watermarking, SynthID, and detecting generated media* Why prompt rewriting matters for video models* Grok Imagine Agent and the rise of video agents* Why language models may unlock better video generation* Robotics, physical AI, and embodied world models* Why Ethan left xAI and shifted focus toward LLMs* Self-managed context, memory, and the next frontier for language modelsEthan He* LinkedIn: https://www.linkedin.com/in/ethanhe42* X: https://x.com/EthanHe_42Timestamps00:00:00 Introduction00:01:25 From NVIDIA Cosmos to xAI00:03:24 Building Grok Imagine from Zero to One00:10:07 How Image and Video Models Are Trained00:18:53 Video Compression, VAEs, and Real-Time Tradeoffs00:22:10 Generative UI, Flipbook, and Neural OS00:32:10 The Cost of Training Large Video Models00:37:04 Distillation, GANs, and Fast Video Inference00:41:21 Audio-Video Generation and Grok Imagine 0.900:48:34 What Makes a World Model?00:55:51 Reference Videos, Long Context, and Video Memory01:00:11 xAI Culture, Research, and First-Principles Building01:09:45 AI Safety, Watermarking, and Prompt Rewriting01:13:10 Video Agents and AI-Assisted Creation01:27:32 Why Language Models Unlock Better Video01:31:15 Robotics, Physical AI, and Embodied World Models01:32:38 Why Ethan Left xAI01:34:16 Self-Managed Context and the Future of LLMs01:38:43 Ethan’s Career Path and Closing ThoughtsTranscriptIntroduction: Ethan He, Latent Space, and the Path to xAISwyx [00:00:00]: We’re here in the studio with Ethan He, most recently of xAI. Welcome.Ethan [00:00:10]: Thank you. Glad being here.Swyx [00:00:11]: We’re also here with Vibhu. you were first coming to us or joining the latent space world because you were working on Kosmos at NVIDIA, and you did a paper. We loved it. you presented it as well, so thank you for doing that.Ethan [00:00:23]: I’ve actually, I also presented the MoEs twice at latent space.Swyx [00:00:29]: How did you actually hear about us? Did we reach out to you? Is that how it worked?Ethan [00:00:33]: No, actually, I-- the community. Like I realized, oh, there is this online community that people talk about AI and also learn from each other through papers every week through the Paperclip. It’s very nice.Ethan [00:00:49]: I learned a lot.Swyx [00:00:49]: I think three years stop. We haven’t stopped even on Christmas and New Years. many weeks I want to stop but it keeps going.Vibhu [00:00:58]: No, that was good. I think you had posted that you worked on a paper, and I was “Oh, very cool. We have Paperclip. Present then.”Vibhu [00:01:04]: But I might have reached out to you after.Swyx [00:01:05]: you-- because it’s an amateur club, right?Swyx [00:01:08]: so it’s very unusual and but we have sometimes paper authors come by and actually explain the paper. Today we just did, the poolside paper, which was apparently very good.Vibhu [00:01:18]: Came out yesterday.Vibhu [00:01:19]: pretty interesting, right? Fully open. They talk about everything, systems. So it’s a good one. We’ll, we’ll recommend people to read it.Swyx [00:01:25]: Bring us up to speed on your transition to xAI, ‘cause I actually don’t eve...
Summary
The Age of Async Agents — Cognition's Walden Yan & OpenInspect's Cole Murray
2w ago01:08:02Summary readyThe new AIEWF website is live! CFPs close in 2 days and we will run our first New Engineer Orientation this weekend, get your tickets booked ASAP as they -will- sell out. Take the AI Engineering Survey and get >$2k in credits and free AIE WF tickets!One of the central tensions in the agents industry is that even while there are major decacorn agent labs like Sierra, Decagon, Notion and Cursor being built up, it is also true that it has never been easier to DIY agents, with a plethora of agent frameworks like LangGraph and Pydantic and Flue, and managed agents from Anthropic and Gemini and Amazon. There has been a wave of companies building their own background agents from Shopify to Stripe to Paradigm to Razorpay, and even Cognition’s friends Ramp have built their own coding agent with other friend Modal.You’d think Cognition might feel a bit threatened, but they’re not - even after all this, they were way oversubscribed for the $1B Series D they just announced:Walden Yan, coiner of context engineering and Chief Product Officer/Cofounder of Cognition, invited OpenInspect’s Cole Murray to talk about why the Devin is in the Details.Full conversation live on the pod today: In retrospect, async agents were the most AGI pilled bet you could make in 2024 - the models weren’t good enough yet to vibecode, and people didn’t trust AI enough to let it rip, nobody (including early Cognition) was sure about the form factors. Now it is obvious:* The first wave of AI coding tools made the developer faster but remain heavily in the loop. Copilor and Cursor’s tab autocomplete are prime examples However, the workflow was still heavily centered around and bottlenecked by the developer’s local workflow: a developer in an IDE, watching the model, accepting or rejecting changes, and pushing code one interaction at a time.* The second wave was local agents: Claude Code, Windsurf, Cursor’s agents pane: first one and increasingly many terminals all running concurrently.* The current Age of Async Agents points to a different future focused more on agent orchestration which drives end-to-end development.According to previous guest Steve Yegge, there are finer-grained 8 levels to agent adoption, but we have collapsed it into three.As Cursor’s Michael Truell put it in The third era of AI software development:Cursor is no longer primarily about writing code. It is about helping developers build the factory that creates their software. This factory is made up of fleets of agents that they interact with as teammates: providing initial direction, equipping them with the tools to work independently, and reviewing their work.The agent should not sit solely inside the developer’s flow. It should be setup to work in the background so that you can give it a task, a repo, a machine, a shell, a browser, tests, memory, and review loops to go do the work somewhere else.In less than a year, the sentiment has shifted from avoiding multi-agent systems:to suggesting approaches that actually work:From coining “context engineering” to building the infrastructure behind Devin’s 7x PR growth and jump from 16% to 80% of commits across Cognition repos, Walden Yan has had a front-row seat to the background-agent shift. In this episode, Cognition co-founder and CPO Walden Yan joins swyx alongside Cole Murray, creator of OpenInspect, to unpack why everyone is building their own Devin, what changed after the December 2025 model inflection, and why “spec to pull request” is now becoming a real production workflow.We go deep on the architecture of background agents: harness-in-the-box vs out-of-the-box, why Devin separates the “brain” from the machine, why repo setup is still one of the hardest problems, why Docker is not always enough, and how full VMs, snapshots, scoped secrets, GitHub bots, Slack integrations, and video-based testing all fit together. Walden and Cole also dig into memory, MCP limitations, multi-agent orchestration, AI code review, SRE auto-triage, PMs shipping code from Slack, Windsurf 2.0, hybrid frontier/sub-frontier systems, and the real failure mode of uncontrolled vibe coding: your codebase regressing to your worst engineer.And as agents eat software… and software eats the world… you can draw the conclusion on what is next:We discuss:* Why the engineering world is waking up to background agents and cloud agents* The December 2025 model inflection that made spec-to-PR workflows practical* Devin’s 7x merged PR growth and rise from 16% to 80% of commits* Why Cole built OpenInspect as an open-source background-agent system* The economics of $20/seat agent products and why monetization is tricky* What Cognition actually sells beyond Devin: infra, onboarding, integrations, and adoption* Harness in the box vs out of the box, and why architecture matters* Why Devin separates the brain from the machine for security and permissions* Repo setup, scoped secrets, Docker Compose, and agent-ready dev environments* Why full VMs matter when agents need to run real applications and test them* Android, macOS, Windows, nested virtualization, and machine-specific agent work* Why testing is much harder than “computer use”* Screenshots, video verification, and the “I know it works” merge moment* GitHub UX, Devin Review, AI reviewers, and agents responding to PR comments* Why MCP alone is not enough for first-class Slack and enterprise integrations* Memory, Knowledge, skills, Claude.md, and why retrieval is still unsolved* Devin’s auto-generated memories and the challenge of memory pruning* Always-on agents as permanent PMs for issues, tickets, and product areas* Sub-agents, meta-Devin management, and what multi-agent systems actually add* Why pure auto-merge vibe coding breaks down after about two weeks* AI code smells, lint rules, reward hacking, and Semgrep for agent-written code* GitAI, inline context, and preserving the “why” behind code changes* Local testing, mock servers, older codebases, and preparing companies for agents* Windsurf 2.0 and the handoff between local foreground agents and cloud background agents* SRE auto-triage, support workflows, and agents as first responders* PMs, marketing, and non-engineers creating pull requests from Slack* AI agent bu...
Summary
🔬ESM: The Bitter Lesson is Coming for Proteins - Alex Rives, BioHub
2w ago01:10:12Summary readyEditor’s note: In our first BioHub pod with Priscilla and Mark they discussed their acquisition of EvoScale, led by Alex Rives, who is now Head of Science at BioHub. With ESM-1 they trained language models on millions of protein sequences drawn from across life, with a simple “next token” objective: predict the amino acids that have been randomly masked out, based on the context of the rest of the sequence. But they soon found that these models also learned biological structure and function, including properties the model had never been explicitly shown AND that this ability scales predictably with compute, leading to ESM2 and ESM3.Today, Alex announced ESMFold 2, an open scientific engine to power prediction, design, and discovery across protein biology.Building on Cryo-EM data (discussed in the CZI pod), ESMFold2 reports state of the art performance on protein interactions, especially antibodies, a critical modality for therapeutics, and evidence that inference time scaling is also working across five targets in cancer and immunology.In a nod to that other famous AI x protein folding project, they are also releasing an atlas of 6.8 billion proteins, and 1.1 billion predicted structures, which you can play around with on their website. We are honored to work with them for this huge release!One of the refrains we’ve heard on the Science pod has been that protein folding, materials design, cellular biology, etc. are very different problems from Language Modeling. They definitely are. Yet Alex Rives and the ESM team at BioHub just released a preprint and model, demonstrating that vanilla BERT-like transformer models trained on sufficiently large and diverse data sets can beat specialized models like AlphaFold3 on some of the hardest protein-related problems. Andrew White had a great segment in our first LS-Science episode that explained how mind blowing AlphaFold2 was when it was released in 2020: it suddenly solved problems on a GPU on your desktop that DESRes had built custom-ASIC supercomputer clusters to solve. John Jumper and Demmis Hassabis received the Nobel Prize in Chemistry for this work.AlphaFold2 took advantage of an very clever observation: if multiple species co-evolve pairs of mutations, this implies that the mutations correspond to parts of the protein that are close in 3d space. This is usually shorthanded as MSAs (multi-sequence alignments), and is the key insight which makes AlphaFold2 so effective.Like other inductive biases, however, it hurts generalization.Scale-pilled before it was coolIf you take a look at the timeline for scaling laws for LLMs and release of structure prediction models, the ESM team notably doubled down on their MSAs-be-damned approach after AlphaFold2 released. This obviously requires a great deal of belief in the scale hypothesis.Why the conviction?ESM developed at a time when many of the scaling laws and the “Bitter Lesson” were proving increasingly correct. AlphaFold2’s wild success must have been both exciting and bitterly disappointing. But using MSAs mean that the model is is dependent on training data that contains MSAs in order to be accurate in a given domain. For things like antibodies that don’t have MSAs to train on, AlphaFold tends to do poorly.ESM takes a different approach: learn the relationship between different proteins by unsupervised training on as much diversity as you can find (sound familiar?) and then correlate that back to structures know from the Protein Data Bank (PDB) and other sources. In other words, a World Model.World Model for proteins“World Model” is a hype term that I define like this:Use unsupervised training to learn abstract patterns from the data:* The abstraction should be semantic - novel constructions represent things that obey the rules of the real world* The abstraction should be compositional - recombining different patterns leads to novel and often valid constructions* The abstraction should support generalization - it predicts things in the real world it wasn’t trained on Once you have a world model, you can attach “heads” to it for downstream tasks: predict properties of a protein, decompose its functional features, or search the representation for proteins that meet design criteria. The two big models BioHub just released under MIT license map directly onto this:* World model → ESMC (a model trained on 2.8 billion sequences)* Structure-prediction head → ESMFold2One of the interesting ways the world model can “predict things” is to generate proteins sequences and then measure the predicted properties, such as binding affinity, in the lab. Alex talks in the episode about validating some of the harder molecules they predicted in the wet-lab. Very cool!Another way is to use mech-interp techniques such as Sparse Auto Encoders (SAEs) to extract semantic features from your model, and then find novel features that predict unknown biology. I won’t spoil this part for you: it was one of the highlights of the episode for me!A cell is a computerWe have all heard that genes are like computer programs, but usually the analogy fizzles after that. Of course genes are transcribed into RNA and RNA is translated into proteins, so genes are programs for building proteins, but that carries the analogy only to “binary digits are programs.” Here’s a better analogy: you can think of the cell nucleus as a storage device / storage controller, the ribosome as a JIT-compiler and runtime, and the semantic features that we learn from our world model via SAEs as functions, proteins as processes that interact together in workflows (signalling pathways) to produce behaviors and outputs (phenotypes). Like functions, the SAE features have a hierarchical composition from local, secondary and tertiary structures (mimicing protein structure), but also motifs that are conceptual, such as membrane integrations, disordered regions and disulfide bonds. As we learn to compose these features we into novel protein designs, we move further towards programmable biology. Alex goes into much more detail about this in the episode, as well as:* Principles for new data collection* BioHub’s vision* Modeling the cellEnjoy!Full Video podcastplease like and subscribe!* X: https://x.com/alexrives* LinkedIn: This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit www.latent.space/subscribe
Summary
Giving Agents Computers — Ivan Burazin, Daytona
3w ago01:10:27Summary readyTake the 2026 AI Engineering Survey and get >$2k in credits and AIE WF tickets!On the product side, everyone is getting Computer - Perplexity, Manus, Cursor, and so on. Meanwhile on the research side, agentic evals like TerminalBench and GDPVal are also assuming computer (Harbor). On both ends, the consolidating LLM OS stack has become a standard toolkit, and Daytona is one of a small set of AI Infra companies that are booming because of it.“The end of localhost” has been Ivan Burazin’s obsession for more than a decade.Something that is all too familiar…Long before agents became the default way people talked about software development, Ivan was already chasing the idea that development should not depend on a fragile local machine. CodeAnywhere, one of the first browser-based IDEs, was an early attempt at that future: move the development environment into the cloud, make setup reproducible, and free developers from the endless “works on my machine” tax.The thesis was directionally right, but the market wasn’t ready yet.However, agents changed that. They do not care about a laptop, desk setup, or favorite editor. They need a computer they can access through an API: something stateful enough to keep working, fast enough to spin up instantly, flexible enough to resize, isolated enough to be safe, and composable enough to run the messy real-world workflows that real software engineering actually requires.Daytona isn’t just selling “sandboxes” in the narrow code-execution sense. It is the latest version of Ivan’s original localhost thesis.In this episode, Daytona’s CEO joins swyx to explain why AI agents need more than code execution boxes: they need composable computers, stateful sandboxes, instant startup, dynamic resources, and infrastructure that can survive workloads going from zero to 100,000 CPUs.We go deep on the new agent compute market: Daytona’s hard pivot from human dev environments to AI sandboxes, the New Year’s Eve MVP that customers begged for, why Daytona runs on bare metal with its own scheduler, how one customer runs almost 850,000 sandboxes a day, and why RL/eval workloads went from 0% to roughly 50% of usage in just months. Ivan also explains why agents need Windows and macOS machines, why CLI may matter more than MCP, why Kubernetes is painful for this workload, and why the future AI cloud may look more like Stripe than AWS.We discuss:* How Daytona grew out of CodeAnywhere, Shift, and the “end of localhost” thesis* Why Daytona pivoted from human dev environments to AI sandboxes* Why agents need composable computers instead of disposable code execution boxes* The New Year’s Eve MVP that customers chased API keys for* Why Daytona chose bare metal, stateful snapshots, and its own scheduler* How Daytona spins up one sandbox in ~60ms and 50,000 sandboxes in ~75 seconds* Why Daytona’s biggest customer runs ~850,000 sandboxes a day* How RL/eval workloads create zero-to-100,000 CPU spikes* Why RL workloads went from 0% to roughly 50% of Daytona usage* Why customers compare Daytona against EKS/GKS and say they’re “never going back”* Why every AI agent may need a computer, including Windows and macOS environments* The Apple licensing constraints that make macOS sandboxes hard* Why CLI gives agents more power than MCP* How open source helps agents integrate Daytona* Why agent-generated PRs may break today’s CI/CD assumptions* Why AI SaaS companies reselling tokens may face a cold shower* Why the AI cloud may look more like Stripe than AWSIvan Burazin* LinkedIn: https://www.linkedin.com/in/ivanburazin* X: https://x.com/ivanburazinDaytona* Website: https://www.daytona.io* X: https://x.com/daytonaioTimestamps* 00:00:00 Hook* 00:01:12 Introduction* 00:03:15 CodeAnywhere, Shift, and the end of localhost* 00:05:58 What Daytona is: composable computers for AI agents* 00:08:07 The pivot from dev environments to AI sandboxes* 00:10:17 The New Year’s Eve MVP and customers begging for API keys* 00:12:56 Bare metal, stateful sandboxes, and Daytona’s scheduler* 00:17:28 60ms startup, 50,000 sandboxes, and 850K daily runs* 00:21:53 Spiky RL/eval workloads and the new agent infra problem* 00:28:12 RL workloads, Kubernetes pain, and dynamic resizing* 00:33:31 Why every AI agent needs a computer* 00:38:48 macOS sandboxes and Apple’s licensing problem* 00:44:28 Why CLI may matter more than MCP* 00:48:11 Open source, GitHub stars, and agent integration* 00:53:11 Git, CI/CD, and agent collaboration bottlenecks* 00:58:15 Founder life and building a 25-person infra company* 01:02:44 AI SaaS, token resale, and API-first business models* 01:06:10 GPU sandboxes, data centers, and compute growth* 01:09:48 Why the AI cloud may look more like Stripe than AWS* 01:11:26 Closing thoughtsTranscriptIntroduction: Daytona, CodeAnywhere, and the End of LocalhostSwyx [00:00:02]: Okay, we’re in the studio with Ivan Burazin, CEO of Daytona. Welcome.Ivan [00:00:07]: Thanks for having me, man.Swyx [00:00:08]: Ivan, you and I go back.Ivan [00:00:10]: Way back.Swyx [00:00:11]: How I don’t even know how, you found, did you reach out or, for Shift.Ivan [00:00:17]: I reached out to you. The reason was you - we were just - we were thinking about I was one of the co-founders of CodeAnywhere, the first browser-based IDE, and so we were thinking a long time of, localhost should die. And you had this article.Swyx [00:00:29]: End of localhost.Ivan [00:00:30]: Then I reached out to you because of that, and then we talked, and I was actually at a different job and learning about I was the head of, developer experience, and you were quite well-versed in that, and I actually reached out to you, among other people, how do we go about that? What are the key things and whatnot at this point in time? And you were nice enough to take the call, and I remember I was late on your call with you.Swyx [00:00:51]: I don’t remember.Ivan [00:00:52]: I remember because I was with my then I’m thinking of a girlfriend or wife at that point in time, I’m not sure. It’s the same person, so that’s great, and I was late ‘cause we were, in, Italy on, vacation, and then I was late for something. I felt so bad, and you were so nice to be, good about.Swyx [00:01:10]: The reason I’m nice is because I’m also late to other people, so it’s like, who’s, who’s without sin here, yeah, so I have to, for those who don’t know, InfoBip Shift, there’s this whole thing that, you did in the past, and, and that was basically one of the inspirations for me starting AI Engineer, which is like, I have to thank you for giving me that push to be like, “Oh, you can, you can build and sell conferences?”Ivan [00:01:34]: I remember you asked you asked me at the beginning to give me advisory shares, and I was so focused on what we were doing, I said no, and I should’ve took the advisory shares. So I’m sorry, dude. But anyway.Swyx [00:01:43]: We’re not, we’re not venture backed.Ivan [00:01:44]: No, it doesn’t matter.Swyx [00:01:45]: It’s Yeah, anyway, so I think what’s impressive about you is that CodeAnywhere is the thing that you’ve been trying to build, and, you kind of put it on hold and then came back after InfoBip. Just give us the story, do you - the story and the origin story, going into Daytona.From CodeAnywhere and Shift to DaytonaIvan [00:02:05]: Sure. Like, really way back, me and my co-founder have been together. I say this, I’ve said this multiple times, it’s like we were married and divorced and married. Some people actually ask me is my co-founder my partner. they thought it literally. It’s not literally, but we have done multiple companies together, and to your point, we had this shift where we went from the CodeAnywhere to the conference called Shift, and then back to, Daytona. We o...
Summary
Railway: The Agent-Native Cloud — Jake Cooper
3w ago01:28:34Summary readyTake the 2026 AI Engineering Survey and get >$2k in credits and AIE WF tickets!This was recorded before Railway suffered a major GCP outage on May 19, despite being a multi-AZ, multi-zone mesh ring, with HA fiber interconnects between their Metal GCP AWS, because workload discoverability was unintentionally still tied to GCP. All has been resolved with a post-mortem.Railway did not start as an AI infrastructure company.It was founded in 2020 years before agents became the default way people thought about deploying software. Jake Cooper, formerly at Bloomberg and Uber, started Railway with a simple obsession: the activation energy to ship something to production should be near zero. Push code, get a URL, iterate. No Docker files, no Kubernetes manifests, no Ansible scripts stacked on Ansible scripts.For years, this was a slow grind. Railway spent its first 18 months hand-acquiring its first 100 users with Jake personally greeting every Discord signup on a second monitor.Today, Railway has raised $124m and is growing very fast. A 35-person team supports 3 million users, adding roughly 100,000 signups a week. Their bare metal data centers have a 3-month payback period vs. renting in the cloud, with 70% margins funding aggressive cloud bursting when needed. The servers they own have actually appreciated in value as RAM prices have climbed basically meaning the value of their hardware now exceeds the capital they've raised.From rebuilding Railway’s network overlay over a weekend to moving the vast majority of workloads onto its own bare metal data centers, Jake Cooper is trying to build a new cloud for an agent-native world. In this episode, Railway’s founder and “conductor” joins swyx and Alessio to unpack why the next era of software infrastructure is not just “Heroku but newer,” what agents need that humans did not, and why the old deployment loop of Git, PRs, CI/CD, and static cloud resources may be heading for a rewrite.We go deep on Railway’s infrastructure stack: own-metal data centers, three-month cloud payback periods, cloud bursting, data center debt, Railpack, Nixpacks, Temporal, feature flags, Central Station, content-addressable filesystems, agent-safe production forks, and why the CLI may become more important than the canvas in an agent world. Jake also shares the founder journey behind Railway, how the company survived losing $500K/month, why it now serves millions of users with only 35 people, and why he believes the pull request is dying.We discuss:* How Railway went from a slow six-year grind to adding 100,000 users a week* How Railway thinks about agents as the next dominant software species* Why agents need version control, observability, compute, storage, and orchestration at 1000x scale* The economics of Railway’s own-metal data centers and three-month payback* How Railway uses cloud bursting while scaling its own infrastructure* Why data center debt can be a better tool than venture debt for infra startups* Central Station, Railway’s internal system for clustering customer feedback and incidents* Why responsible disclosure and over-communication matter for platforms* Why feature flags, progressive rollouts, and shadow traffic are essential for agents* Temporal’s strengths, pain points, and why workflows matter for agents* Railpack, Nixpacks, Nix, and lazy-loaded content-addressable filesystems* Why “cattle, not pets” may change if you can clone the pets* Why Railway is building a new cloud from scratch instead of copying hyperscalers* The solo founder path, focus, writing, and how Jake thinks about company buildingRailway:* Website: https://railway.com/* X: https://x.com/RailwayJake Cooper:* LinkedIn: https://www.linkedin.com/in/thejakecooper/* X: https://x.com/JustJakeTimestamps00:00:00 Introduction: What Is Railway?00:02:07 Jake’s Path to Railway00:06:13 Railway’s Six-Year Growth Story00:08:52 Rebuilding the Business After the Free Tier00:11:17 Agents as the Next Software Platform00:13:29 Railway’s Infrastructure Philosophy00:15:42 Bare Metal, Cloud Economics, and the Compute Crunch00:17:22 Cloud Bursting and Five-Cloud Networking00:20:20 Data Center Debt and Infra Financing00:23:31 Data Centers in Space00:25:24 What Agents Need From Infrastructure00:28:24 CLIs, Canvas, and Agent-Native UX00:35:15 Central Station, Incidents, and Responsible Disclosure00:40:30 Safe Rollouts, SRE Agents, and Production Forks00:45:00 AI SRE, Specs, Code, and Tests00:48:24 Self-Replicating Infrastructure and the New Serverless00:53:18 Heroku, Temporal, and Workflow Engines01:04:07 Railpack, Nixpacks, and Lazy-Loaded Filesystems01:06:01 Coding Agents, Token Spend, and Roadmap Acceleration01:10:56 The Pull Request Is Dying01:12:28 Feature Flags and the Agent-Era SDLC01:16:15 Cattle, Pets, and Cloning Machines01:19:29 Solo Founder Lessons01:24:12 Focus, GPUs, and Building a New Cloud01:28:20 Closing ThoughtsTranscriptAlessio [00:00:00]: Hey, everyone. Welcome to the Latent Space Podcast. This is Alessio, founder of Kernel Labs, and I’m joined by Swyx, editor of Latent Space.Swyx [00:00:10]: Hey, hey, hey. Today we’re in the studio with Jake Cooper of Railway.Alessio [00:00:14]: Conductor of Railway.Swyx [00:00:15]: Conductor at Railway. Yeah.Alessio [00:00:16]: Choo-choo.Swyx [00:00:17]: Do you actually have that anywhere, like on your business card?Jake [00:00:20]: We call some of our volunteer moderators conductors. I don’t have a business card. We’re not that big yet. At some point I will. I got handed a nice business card from the Supermicro folks, and I was like, “Damn, this is pretty official.”Swyx [00:00:30]: Business cards are coming back.Jake [00:00:32]: They’re cool. They’re hip. The conductor thing is good. We’re trying to figure out what we want to call each other internally. Some people think it’s super cringe and say, “You don’t need a name for people internally.” Some people want to call each other something. We still don’t have a really good one.Jake [00:00:55]: We’ve got New Railcrews, Trainiacs. Nothing has stuck yet.Swyx [00:01:00]: I like Trainiac. Trainiac sounds good. Railwayians. For those who don’t know, what is Railway? Let’s give people a crisp definition up front.Jake [00:01:09]: Railway is the easiest way to ship anything. You go to the canvas, or you talk with Claude, and you say, “Deploy a Postgres instance, deploy my GitHub repository, run this code,” and you’re off to the races.Swyx [00:01:22]: You’ve got a nice animation on the landing page.Jake [00:01:24]: Thank you. None of my work, by the way. They don’t let me touch the design stuff anymore.Jake [00:01:25]: We want to make it trivially easy not just to deploy things, but to evolve applications over time. Most tooling right now stacks entropy on top of entropy: Docker, Kubernetes, Ansible scripts, and all these other things. If we can version all of your software and keep track of all the changes, then we can make it trivial to clone environments, fork into a parallel universe, get copies of production data, get copies of any services, make changes, validate them, and collapse them back in without reproducing everything across a staging environment.The Railway Origin Story: From Uber Systems to a New CloudSwyx [00:02:07]: I was looking at your background: Bloomberg, Uber. Nothing immediately stands out as, “This guy is going to found the next great platform as a service.” What prepared you for Railway?Jake [00:02:21]: It was curiosity to keep going deeper. I started out on front-end stuff, working on Wolfram Mathematica and porting it over. Then I briefly moved to Bloomberg, then toward Uber and distributed systems, taking the Jump Bikes systems and moving them to a distributed system built on top of Cadence, the pre-Temporal Temporal.Swyx [00:02:44]: Which, by the way, I’m happy to talk about, pros and cons.Jake [00:02:48]: Totally.Swyx [00:02:51]: But let’s do the Railway...
Summary
The Autonomous Drone Tech Stack & Economics of Drones — Yaroslav Azhnyuk, The Fourth Law & Guest Host Noah Smith, Noahpinion
3w ago01:59:28Summary readyThe future of war has been evolving before our eyes in Ukraine, yet the west still plans to fight the last war. In this special episode, guest host Noah Smith (@noahpinion) and Brandon Anderson sit down with Yaroslav Azhnyuk (@YaroslavAzhnyuk), a serial tech founder who went from building PetCube to founding The Fourth Law, one of the world’s most advanced AI-guided drone companies. Over two hours we cover the technology, tactics, and geopolitics of drone warfare, and why the modern battlefield has already left the West behind:* Yaroslav’s personal history and the Ukraine war [00:01:04 – 00:14:01]* The modern drone tech stack: why FPV drones are the new god of war, the future of the rifleman, fiber optic vs. AI, five levels of autonomy, and the eight dimensions of the autonomous battlefield [00:14:01 – 01:05:13]* The geopolitics and economics of drones: China’s manufacturing advantage, the drone race, Western defense readiness, countermeasures, and why the gap is widening [01:05:13 – 01:58:57]For those looking for Noah Smith’s commentary, it really gets going around the 00:51:31 mark.Yaroslav Azhnyuk / The Fourth Law:* X: https://x.com/YaroslavAzhnyuk* LinkedIn: https://www.linkedin.com/in/yaroslavazhnyuk/* The Fourth Law: https://thefourthlaw.aiNoah Smith:* Substack: Noah Smith * X: https://x.com/noahpinionTimestamps00:00:00 Cold Open: China’s 4 Billion Drones and the Cameras-to-Explosives Pipeline00:01:04 Introduction: Brandon, Noah Smith, and Yaroslav Azhnyuk00:05:41 From Tech Entrepreneur to Defense: PetCube, Brave One, and the D3 Fund00:10:42 The Ethics of Building Weapons: Dual-Use Technology and the Wolf at the Door00:14:01 The Tech Stack: Cameras, Autonomy Modules, Interceptors, and a Semiconductor Fab00:18:47 Fiber Optic vs. AI: The Radio Horizon Problem and $32/km Cable00:25:32 FPV Drones: The New God of War — 70–80% of Frontline Casualties00:28:28 The Five Levels of Drone Autonomy: From Terminal Guidance to Full Autonomy00:41:37 The Eight Dimensions of the Autonomous Battlefield00:45:32 AI Safety and the Morality of Autonomous Weapons00:51:31 The End of the Rifleman? Noah’s 2013 Prediction vs. Battlefield Reality01:05:13 China’s Manufacturing Advantage and Western Vulnerabilities01:24:21 Policy Advice for Western Defense: Defense Valley and the Widening Gap01:32:54 The Drone Race: Who’s Ahead, Category by Category01:41:57 Countermeasures: Shotguns, Jammers, Lasers, and Fishnets01:58:19 The Wedding and Final Takeaway: Be Prepared for WarTranscriptCold Open: China, FPV Drones, and the New Warning SignYaroslav [00:00:00]: Think about this. Last year, Ukraine produced 4 million FPV drones. Ukraine is not the most industrious nation in the world. China can produce 4 billion of these FPV drones.Noah [00:00:10]: Would you say that right now China is now the supreme conventional military power on Earth, given its ability to manufacture and deploy drones in the quantity and quality that you just described?Yaroslav [00:00:20]: I don’t think we have all the information to claim that but we cannot count it out, and that alone should be a big warning sign. As I say, at some point in my life I went from making cameras that fling treats to pets to cameras that fling explosives to the occupiers. So that’s the short story. And when you think about what your nation, what your patriots are going through, you realize that’s the only morally right thing to do is to fight back, and it is immoral not to fight back, and then the choice becomes very clear.Introduction: Yaroslav Azhnyuk, Petcube, and the Last Flight into KyivBrandon [00:01:04]: Welcome to Latent Space. I’m Brandon. I normally do science podcasts, but today we’re going to do something a little bit different. I’m joined by Noah Smith of Noahpinion on Substack and Twitter. And he has lots of interesting things to say about drones. And as a guest, we have Yaroslav Azhnyuk, founder of The Fourth Law and several other, drone-related startups. To get started, it is February 23rd, 2022. You are running a pet startup. You’re connecting pets with their owners. Let’s go in just a little bit of background. How did you get started in tech, and what were you working on before the Ukrainian war started?Yaroslav [00:01:50]: Good to be here. Thank you. On February 23rd, late in the evening, 11:00 PM Kyiv time, my wife and I landed in Kyiv. Actually, then she was a fiance. We came from Lviv, where we were looking at a church, where our wedding should have taken place. And we got into this cab ride from the airport to our home, and the driver was like, “You crazy. Like, everyone’s leaving Kyiv. Why do you come?” We’re like, “What? Nothing’s going to happen. Dude, chill.” And then obviously, eight minutes later, or eight hours later, the bombs fell in the city. It was quite surreal. We probably landed on the last flight that landed in Kyiv, or one of those last flights. My background, I’m a tech guy. Studied applied mathematics in Kyiv Polytechnics, born and raised in Kyiv. My parents are old PhDs from academia, and grandparents too. Like, everything, from linguistics to nuclear physics. And I’m an entrepreneur, so I’ve built a bunch of companies. Petcube is the one you were referencing. So I lived in San Francisco 2014 to 2020, building Petcube, which is one of the leading, pet device companies in the world, selling lots of pet cameras. And then, yeah, as I say, at some point in my life I went from making cameras that fling treats to pets to cameras that fling explosives to the occupiers. So that’s the short story.February 24th: Leaving Kyiv as the Invasion BeginsNoah [00:03:28]: February 24th, I guess a few hours after you, go to check out your wedding chapel, what do you do?Yaroslav [00:03:37]: We had a plan for this situation. So my parents and family live in Kyiv, and we’re like, “Okay, this has actually started. The worst has, come true.” And so we basically packed our belongings and got in the car and spent 17 hours driving west. And that was pretty sure most people in our audience watched at least one apocalyptic movie in their life, so that was exactly like that. Like, felt exactly like that. Missiles are falling. Like, there was smoke in Kyiv. Like, my dad and I went, like, to central part of the cities. It’s probably, likeYaroslav [00:04:20]: 800 meters from presidential office, to pick some stuff up at his workplace. Because he’s, like, the head of an academic institution, so he had to get some of the things with him. And super surreal. Like, the streets are empty. Like, the gas stations are out of gas. Like, we found some gas station. We didn’t have, like, spare canisters with us, so we’re like, We figured out, like, the car was diesel, so like, we figured out, if it’s diesel, you can actually store it in plastic, canisters, and we bought some window wash for the cars. We poured it out of the canisters, and we poured the diesel into that. Yeah, so it was like that. And then, like, helping friends get out, like my friend and his dog. Like, we found Like, my brother was also, like, riding in a separate car. We found a place for my friend who didn’t have a car. It was like, yeah, it was like, totally surreal. And we didn’t know of course, and you didn’t know this will last for so long. You didn’t know whether Ukraine will be able to defend Kyiv. And it was like, yeah, very little information and very little insight into future.From Pet Cameras to Defense Tech: Building for Ukraine and the Free WorldNoah [00:05:42]: What are your thoughts with regards to how do you, defend, Ukraine? So you eventually start building drones Like, what is the process to get from there from where you were building, devices that connect owners with pets to building drones, and what other things did you do to help the war effort in the process?Yaroslav [00:06:07]: It’s definitely non-trivial, right? Like, I didn’t go, to I didn’t get any, like, military education when I was a student. Like, normally, in Ukraine, you would, you would go to like, this military school even if you’re getting higher education in any other, sphere. I decided to skip that which is like, an unusual way to go. And I never thought that I will be somehow engaged in a war effort. Like, what is war? Of course, wars are over. It’s the end of history. So one thing you got to understand about, like, many Ukrainians and like, I guess, it’s also true about most of the people I met here in the US, that your who you are in terms of your nationality is a big part of your identity. So when t...
Summary