Explore every episode of the podcast Latent Space: The AI Engineer Podcast

From undergraduate research seminars at Princeton to winning Best Paper award at NeurIPS 2025, Kevin Wang, Ishaan Javali, Michał Bortkiewicz, Tomasz Trzcinski, Benjamin Eysenbach defied conventional wisdom by scaling reinforcement learning networks to 1,000 layers deep—unlocking performance gains that the RL community thought impossible. We caught up with the team live at NeurIPS to dig into the story behind RL1000: why deep networks have worked in language and vision but failed in RL for over a decade (spoiler: it's not just about depth, it's about the objective), how they discovered that self-supervised RL (learning representations of states, actions, and future states via contrastive learning) scales where value-based methods collapse, the critical architectural tricks that made it work (residual connections, layer normalization, and a shift from regression to classification), why scaling depth is more parameter-efficient than scaling width (linear vs. quadratic growth), how Jax and GPU-accelerated environments let them collect hundreds of millions of transitions in hours (the data abundance that unlocked scaling in the first place), the "critical depth" phenomenon where performance doesn't just improve—it multiplies once you cross 15M+ transitions and add the right architectural components, why this isn't just "make networks bigger" but a fundamental shift in RL objectives (their code doesn't have a line saying "maximize rewards"—it's pure self-supervised representation learning), how deep teacher, shallow student distillation could unlock deployment at scale (train frontier capabilities with 1000 layers, distill down to efficient inference models), the robotics implications (goal-conditioned RL without human supervision or demonstrations, scaling architecture instead of scaling manual data collection), and their thesis that RL is finally ready to scale like language and vision—not by throwing compute at value functions, but by borrowing the self-supervised, representation-learning paradigms that made the rest of deep learning work. We discuss: The self-supervised RL objective: instead of learning value functions (noisy, biased, spurious), they learn representations where states along the same trajectory are pushed together, states along different trajectories are pushed apart—turning RL into a classification problem Why naive scaling failed: doubling depth degraded performance, doubling again with residual connections and layer norm suddenly skyrocketed performance in one environment—unlocking the "critical depth" phenomenon Scaling depth vs. width: depth grows parameters linearly, width grows quadratically—depth is more parameter-efficient and sample-efficient for the same performance The Jax + GPU-accelerated environments unlock: collecting thousands of trajectories in parallel meant data wasn't the bottleneck, and crossing 15M+ transitions was when deep networks really paid off The blurring of RL and self-supervised learning: their code doesn't maximize rewards directly, it's an actor-critic goal-conditioned RL algorithm, but the learning burden shifts to classification (cross-entropy loss, representation learning) instead of TD error regression Why scaling batch size unlocks at depth: traditional RL doesn't benefit from larger batches because networks are too small to exploit the signal, but once you scale depth, batch size becomes another effective scaling dimension — RL1000 Team (Princeton) 1000 Layer Networks for Self-Supervised RL: Scaling Depth Can Enable New Goal-Reaching Capabilities: https://openreview.net/forum?id=s0JVsx3bx1 Chapters 00:00:00 Introduction: Best Paper Award and NeurIPS Poster Experience 00:01:11 Team Introductions and Princeton Research Origins 00:03:35 The Deep Learning Anomaly: Why RL Stayed Shallow 00:04:35 Self-Supervised RL: A Different Approach to Scaling 00:05:13 The Breakthrough Moment: Residual Connections and Critical Depth 00:07:15 Architectural Choices: Borrowing from ResNets and Avoiding Vanishing Gradients 00:07:50 Clarifying the Paper: Not Just Big Networks, But Different Objectives 00:08:46 Blurring the Lines: RL Meets Self-Supervised Learning 00:09:44 From TD Errors to Classification: Why This Objective Scales 00:11:06 Architecture Details: Building on Braw and SymbaFowl 00:12:05 Robotics Applications: Goal-Conditioned RL Without Human Supervision 00:13:15 Efficiency Trade-offs: Depth vs Width and Parameter Scaling 00:15:48 JAX and GPU-Accelerated Environments: The Data Infrastructure 00:18:05 World Models and Next State Classification 00:22:37 Unlocking Batch Size Scaling Through Network Capacity 00:24:10 Compute Requirements: State-of-the-Art on a Single GPU 00:21:02 Future Directions: Distillation, VLMs, and Hierarchical Planning 00:27:15 Closing Thoughts: Challenging Conventional Wisdom in RL Scaling

From creating SWE-bench in a Princeton basement to shipping CodeClash, SWE-bench Multimodal, and SWE-bench Multilingual, John Yang has spent the last year and a half watching his benchmark become the de facto standard for evaluating AI coding agents—trusted by Cognition (Devin), OpenAI, Anthropic, and every major lab racing to solve software engineering at scale. We caught up with John live at NeurIPS 2025 to dig into the state of code evals heading into 2026: why SWE-bench went from ignored (October 2023) to the industry standard after Devin's launch (and how Walden emailed him two weeks before the big reveal), how the benchmark evolved from Django-heavy to nine languages across 40 repos (JavaScript, Rust, Java, C, Ruby), why unit tests as verification are limiting and long-running agent tournaments might be the future (CodeClash: agents maintain codebases, compete in arenas, and iterate over multiple rounds), the proliferation of SWE-bench variants (SWE-bench Pro, SWE-bench Live, SWE-Efficiency, AlgoTune, SciCode) and how benchmark authors are now justifying their splits with curation techniques instead of just "more repos," why Tau-bench's "impossible tasks" controversy is actually a feature not a bug (intentionally including impossible tasks flags cheating), the tension between long autonomy (5-hour runs) vs. interactivity (Cognition's emphasis on fast back-and-forth), how Terminal-bench unlocked creativity by letting PhD students and non-coders design environments beyond GitHub issues and PRs, the academic data problem (companies like Cognition and Cursor have rich user interaction data, academics need user simulators or compelling products like LMArena to get similar signal), and his vision for CodeClash as a testbed for human-AI collaboration—freeze model capability, vary the collaboration setup (solo agent, multi-agent, human+agent), and measure how interaction patterns change as models climb the ladder from code completion to full codebase reasoning. We discuss: John's path: Princeton → SWE-bench (October 2023) → Stanford PhD with Diyi Yang and the Iris Group, focusing on code evals, human-AI collaboration, and long-running agent benchmarks The SWE-bench origin story: released October 2023, mostly ignored until Cognition's Devin launch kicked off the arms race (Walden emailed John two weeks before: "we have a good number") SWE-bench Verified: the curated, high-quality split that became the standard for serious evals SWE-bench Multimodal and Multilingual: nine languages (JavaScript, Rust, Java, C, Ruby) across 40 repos, moving beyond the Django-heavy original distribution The SWE-bench Pro controversy: independent authors used the "SWE-bench" name without John's blessing, but he's okay with it ("congrats to them, it's a great benchmark") CodeClash: John's new benchmark for long-horizon development—agents maintain their own codebases, edit and improve them each round, then compete in arenas (programming games like Halite, economic tasks like GDP optimization) SWE-Efficiency (Jeffrey Maugh, John's high school classmate): optimize code for speed without changing behavior (parallelization, SIMD operations) AlgoTune, SciCode, Terminal-bench, Tau-bench, SecBench, SRE-bench: the Cambrian explosion of code evals, each diving into different domains (security, SRE, science, user simulation) The Tau-bench "impossible tasks" debate: some tasks are underspecified or impossible, but John thinks that's actually a feature (flags cheating if you score above 75%) Cognition's research focus: codebase understanding (retrieval++), helping humans understand their own codebases, and automatic context engineering for LLMs (research sub-agents) The vision: CodeClash as a testbed for human-AI collaboration—vary the setup (solo agent, multi-agent, human+agent), freeze model capability, and measure how interaction changes as models improve — John Yang SWE-bench: https://www.swebench.com X: https://x.com/jyangballin Chapters 00:00:00 Introduction: John Yang on SWE-bench and Code Evaluations 00:00:31 SWE-bench Origins and Devon's Impact on the Coding Agent Arms Race 00:01:09 SWE-bench Ecosystem: Verified, Pro, Multimodal, and Multilingual Variants 00:02:17 Moving Beyond Django: Diversifying Code Evaluation Repositories 00:03:08 Code Clash: Long-Horizon Development Through Programming Tournaments 00:04:41 From Halite to Economic Value: Designing Competitive Coding Arenas 00:06:04 Ofir's Lab: SWE-ficiency, AlgoTune, and SciCode for Scientific Computing 00:07:52 The Benchmark Landscape: TAU-bench, Terminal-bench, and User Simulation 00:09:20 The Impossible Task Debate: Refusals, Ambiguity, and Benchmark Integrity 00:12:32 The Future of Code Evals: Long Autonomy vs Human-AI Collaboration 00:14:37 Call to Action: User Interaction Data and Codebase Understanding Research

From building LMArena in a Berkeley basement to raising $100M and becoming the de facto leaderboard for frontier AI, Anastasios Angelopoulos returns to Latent Space to recap 2025 in one of the most influential platforms in AI—trusted by millions of users, every major lab, and the entire industry to answer one question: which model is actually best for real-world use cases? We caught up with Anastasios live at NeurIPS 2025 to dig into the origin story (spoiler: it started as an academic project incubated by Anjney Midha at a16z, who formed an entity and gave grants before they even committed to starting a company), why they decided to spin out instead of staying academic or nonprofit (the only way to scale was to build a company), how they're spending that $100M (inference costs, React migration off Gradio, and hiring world-class talent across ML, product, and go-to-market), the leaderboard delusion controversy and why their response demolished the paper's claims (factual errors, misrepresentation of open vs. closed source sampling, and ignoring the transparency of preview testing that the community loves), why platform integrity comes first (the public leaderboard is a charity, not a pay-to-play system—models can't pay to get on, can't pay to get off, and scores reflect millions of real votes), how they're expanding into occupational verticals (medicine, legal, finance, creative marketing) and multimodal arenas (video coming soon), why consumer retention is earned every single day (sign-in and persistent history were the unlock, but users are fickle and can leave at any moment), the Gemini Nano Banana moment that changed Google's market share overnight (and why multimodal models are becoming economically critical for marketing, design, and AI-for-science), how they're thinking about agents and harnesses (Code Arena evaluates models, but maybe it should evaluate full agents like Devin), and his vision for Arena as the central evaluation platform that provides the North Star for the industry—constantly fresh, immune to overfitting, and grounded in millions of real-world conversations from real users. We discuss: The $100M raise: use of funds is primarily inference costs (funding free usage for tens of millions of monthly conversations), React migration off Gradio (custom loading icons, better developer hiring, more flexibility), and hiring world-class talent The scale: 250M+ conversations on the platform, tens of millions per month, 25% of users do software for a living, and half of users are now logged in The leaderboard illusion controversy: Cohere researchers claimed undisclosed private testing created inequities, but Arena's response demolished the paper's factual errors (misrepresented open vs. closed source sampling, ignored transparency of preview testing that the community loves) Why preview testing is loved by the community: secret codenames (Gemini Nano Banana, named after PM Naina's nickname), early access to unreleased models, and the thrill of being first to vote on frontier capabilities The Nano Banana moment: changed Google's market share overnight, billions of dollars in stock movement, and validated that multimodal models (image generation, video) are economically critical for marketing, design, and AI-for-science New categories: occupational and expert arenas (medicine, legal, finance, creative marketing), Code Arena, and video arena coming soon Consumer retention: sign-in and persistent history were the unlock, but users are fickle and earned every single day—"every user is earned, they can leave at any moment" — Anastasios Angelopoulos Arena: https://lmarena.ai X: https://x.com/arena Chapters 00:00:00 Introduction: Anastasios from Arena and the LM Arena Journey 00:01:36 The Anjney Midha Incubation: From Berkeley Basement to Startup 00:02:47 The Decision to Start a Company: Scaling Beyond Academia 00:03:38 The $100M Raise: Use of Funds and Platform Economics 00:05:10 Arena's User Base: 5M+ Users and Diverse Demographics 00:06:02 The Competitive Landscape: Artificial Analysis, AI.xyz, and Arena's Differentiation 00:08:12 Educational Value and Learning from the Community 00:08:41 Technical Migration: From Gradio to React and Platform Evolution 00:10:18 Leaderboard Delusion Paper: Addressing Critiques and Maintaining Integrity 00:12:29 Nano Banana Moment: How Preview Models Create Market Impact 00:13:41 Multimodal AI and Image Generation: From Skepticism to Economic Value 00:15:37 Core Principles: Platform Integrity and the Public Leaderboard as Charity 00:18:29 Future Roadmap: Expert Categories, Multimodal, Video, and Occupational Verticals 00:19:10 API Strategy and Focus: Doing One Thing Well 00:19:51 Community Management and Retention: Sign-In, History, and Daily Value 00:22:21 Partnerships and Agent Evaluation: From Devon to Full-Featured Harnesses 00:21:49 Hiring and Building a High-Performance Team

From pre-training data curation to shipping GPT-4o, o1, o3, and now GPT-5 thinking and the shopping model, Josh McGrath has lived through the full arc of OpenAI's post-training evolution—from the PPO vs DPO debates of 2023 to today's RLVR era, where the real innovation isn't optimization methods but data quality, signal trust, and token efficiency. We sat down with Josh at NeurIPS 2025 to dig into the state of post-training heading into 2026: why RLHF and RLVR are both just policy gradient methods (the difference is the input data, not the math), how GRPO from DeepSeek Math was underappreciated as a shift toward more trustworthy reward signals (math answers you can verify vs. human preference you can't), why token efficiency matters more than wall-clock time (GPT-5 to 5.1 bumped evals and slashed tokens), how Codex has changed his workflow so much he feels "trapped" by 40-minute design sessions followed by 15-minute agent sprints, the infrastructure chaos of scaling RL ("way more moving parts than pre-training"), why long context will keep climbing but agents + graph walks might matter more than 10M-token windows, the shopping model as a test bed for interruptability and chain-of-thought transparency, why personality toggles (Anton vs Clippy) are a real differentiator users care about, and his thesis that the education system isn't producing enough people who can do both distributed systems and ML research—the exact skill set required to push the frontier when the bottleneck moves every few weeks. We discuss: Josh's path: pre-training data curation → post-training researcher at OpenAI, shipping GPT-4o, o1, o3, GPT-5 thinking, and the shopping model Why he switched from pre-training to post-training: "Do I want to make 3% compute efficiency wins, or change behavior by 40%?" The RL infrastructure challenge: way more moving parts than pre-training (tasks, grading setups, external partners), and why babysitting runs at 12:30am means jumping into unfamiliar code constantly How Codex has changed his workflow: 40-minute design sessions compressed into 15-minute agent sprints, and the strange "trapped" feeling of waiting for the agent to finish The RLHF vs RLVR debate: both are policy gradient methods, the real difference is data quality and signal trust (human preference vs. verifiable correctness) Why GRPO (from DeepSeek Math) was underappreciated: not just an optimization trick, but a shift toward reward signals you can actually trust (math answers over human vibes) The token efficiency revolution: GPT-5 to 5.1 bumped evals and slashed tokens, and why thinking in tokens (not wall-clock time) unlocks better tool-calling and agent workflows Personality toggles: Anton (tool, no warmth) vs Clippy (friendly, helpful), and why Josh uses custom instructions to make his model "just a tool" The router problem: having a router at the top (GPT-5 thinking vs non-thinking) and an implicit router (thinking effort slider) creates weird bumps, and why the abstractions will eventually merge Long context: climbing Graph Blocks evals, the dream of 10M+ token windows, and why agents + graph walks might matter more than raw context length Why the education system isn't producing enough people who can do both distributed systems and ML research, and why that's the bottleneck for frontier labs The 2026 vision: neither pre-training nor post-training is dead, we're in the fog of war, and the bottleneck will keep moving (so emotional stability helps) — Josh McGrath OpenAI: https://openai.com https://x.com/j_mcgraph Chapters 00:00:00 Introduction: Josh McGrath on Post-Training at OpenAI 00:04:37 The Shopping Model: Black Friday Launch and Interruptability 00:07:11 Model Personality and the Anton vs Clippy Divide 00:08:26 Beyond PPO vs DPO: The Data Quality Spectrum in RL 00:01:40 Infrastructure Challenges: Why Post-Training RL is Harder Than Pre-Training 00:13:12 Token Efficiency: The 2D Plot That Matters Most 00:03:45 Codex Max and the Flow Problem: 40 Minutes of Planning, 15 Minutes of Waiting 00:17:29 Long Context and Graph Blocks: Climbing Toward Perfect Context 00:21:23 The ML-Systems Hybrid: What's Hard to Hire For 00:24:50 Pre-Training Isn't Dead: Living Through Technological Revolution

From Berkeley robotics and OpenAI's 2017 Dota-era internship to shipping RL breakthroughs on GPT-4o, o1, and o3, and now leading model development at Cursor, Ashvin Nair has done it all. We caught up with Ashvin at NeurIPS 2025 to dig into the inside story of OpenAI's reasoning team (spoiler: it went from a dozen people to 300+), why IOI Gold felt reachable in 2022 but somehow didn't change the world when o1 actually achieved it, how RL doesn't generalize beyond the training distribution (and why that means you need to bring economically useful tasks into distribution by co-designing products and models), the deeper lessons from the RL research era (2017–2022) and why most of it didn't pan out because the community overfitted to benchmarks, how Cursor is uniquely positioned to do continual learning at scale with policy updates every two hours and product-model co-design that keeps engineers in the loop instead of context-switching into ADHD hell, and his bet that the next paradigm shift is continual learning with infinite memory—where models experience something once (a bug, a mistake, a user pattern) and never forget it, storing millions of deployment tokens in weights without overloading capacity. We discuss: Ashvin's path: Berkeley robotics PhD → OpenAI 2017 intern (Dota era) → o1/o3 reasoning team → Cursor ML lead in three months Why robotics people are the most grounded at NeurIPS (they work with the real world) and simulation people are the most unhinged (Lex Fridman's take) The IOI Gold paradox: "If you told me we'd achieve IOI Gold in 2022, I'd assume we could all go on vacation—AI solved, no point working anymore. But life is still the same." The RL research era (2017–2022) and why most of it didn't pan out: overfitting to benchmarks, too many implicit knobs to tune, and the community rewarding complex ideas over simple ones that generalize Inside the o1 origin story: a dozen people, conviction from Ilya and Jakob Pachocki that RL would work, small-scale prototypes producing "surprisingly accurate reasoning traces" on math, and first-principles belief that scaled The reasoning team grew from ~12 to 300+ people as o1 became a product and safety, tooling, and deployment scaled up Why Cursor is uniquely positioned for continual learning: policy updates every two hours (online RL on tab), product and ML sitting next to each other, and the entire software engineering workflow (code, logs, debugging, DataDog) living in the product Composer as the start of product-model co-design: smart enough to use, fast enough to stay in the loop, and built by a 20–25 person ML team with high-taste co-founders who code daily The next paradigm shift: continual learning with infinite memory—models that experience something once (a bug, a user mistake) and store it in weights forever, learning from millions of deployment tokens without overloading capacity (trillions of pretraining tokens = plenty of room) Why off-policy RL is unstable (Ashvin's favorite interview question) and why Cursor does two-day work trials instead of whiteboard interviews The vision: automate software engineering as a process (not just answering prompts), co-design products so the entire workflow (write code, check logs, debug, iterate) is in-distribution for RL, and make models that never make the same mistake twice — Ashvin Nair Cursor: https://cursor.com X: https://x.com/ashvinnair_ Chapters 00:00:00 Introduction: From Robotics to Cursor via OpenAI 00:01:58 The Robotics to LLM Agent Transition: Why Code Won 00:09:11 RL Research Winter and Academic Overfitting 00:11:45 The Scaling Era and Moving Goalposts: IOI Gold Doesn't Mean AGI 00:21:30 OpenAI's Reasoning Journey: From Codex to O1 00:20:03 The Blip: Thanksgiving 2023 and OpenAI Governance 00:22:39 RL for Reasoning: The O-Series Conviction and Scaling 00:25:47 O1 to O3: Smooth Internal Progress vs External Hype Cycles 00:33:07 Why Cursor: Co-Designing Products and Models for Real Work 00:34:14 Composer and the Future: Online Learning Every Two Hours 00:35:15 Continual Learning: The Missing Paradigm Shift 00:44:00 Hiring at Cursor and Why Off-Policy RL is Unstable

From investing through the modern data stack era (DBT, Fivetran, and the analytics explosion) to now investing at the frontier of AI infrastructure and applications at Amplify Partners, Sarah Catanzaro has spent years at the intersection of data, compute, and intelligence—watching categories emerge, merge, and occasionally disappoint. We caught up with Sarah live at NeurIPS 2025 to dig into the state of AI startups heading into 2026: why $100M+ seed rounds with no near-term roadmap are now the norm (and why that terrifies her), what the DBT-Fivetran merger really signals about the modern data stack (spoiler: it's not dead, just ready for IPO), how frontier labs are using DBT and Fivetran to manage training data and agent analytics at scale, why data catalogs failed as standalone products but might succeed as metadata services for agents, the consumerization of AI and why personalization (memory, continual learning, K-factor) is the 2026 unlock for retention and growth, why she thinks RL environments are a fad and real-world logs beat synthetic clones every time, and her thesis for the most exciting AI startups: companies that marry hard research problems (RAG, rule-following, continual learning) with killer applications that were simply impossible before. We discuss: The DBT-Fivetran merger: not the death of the modern data stack, but a path to IPO scale (targeting $600M+ combined revenue) and a signal that both companies were already winning their categories How frontier labs use data infrastructure: DBT and Fivetran for training data curation, agent analytics, and managing increasingly complex interactions—plus the rise of transactional databases (RocksDB) and efficient data loading (Vortex) for GPU-bound workloads Why data catalogs failed: built for humans when they should have been built for machines, focused on discoverability when the real opportunity was governance, and ultimately subsumed as features inside Snowflake, DBT, and Fivetran The $100M+ seed phenomenon: raising massive rounds at billion-dollar valuations with no 6-month roadmap, seven-day decision windows, and founders optimizing for signal ("we're a unicorn") over partnership or dilution discipline Why world models are overhyped but underspecified: three competing definitions, unclear generalization across use cases (video games ≠ robotics ≠ autonomous driving), and a research problem masquerading as a product category The 2026 theme: consumerization of AI via personalization—memory management, continual learning, and solving retention/churn by making products learn skills, preferences, and adapt as the world changes (not just storing facts in cursor rules) Why RL environments are a fad: labs are paying 7–8 figures for synthetic clones when real-world logs, traces, and user activity (à la Cursor) are richer, cheaper, and more generalizable Sarah's investment thesis: research-driven applications that solve hard technical problems (RAG for Harvey, rule-following for Sierra, continual learning for the next killer app) and unlock experiences that were impossible before Infrastructure bets: memory, continual learning, stateful inference, and the systems challenges of loading/unloading personalized weights at scale Why K-factor and growth fundamentals matter again: AI felt magical in 2023–2024, but as the magic fades, retention and virality are back—and most AI founders have never heard of K-factor — Sarah Catanzaro X: https://x.com/sarahcat21 Amplify Partners: https://amplifypartners.com/ Where to find Latent Space X: https://x.com/latentspacepod Substack: https://www.latent.space/ Chapters 00:00:00 Introduction: Sarah Catanzaro's Journey from Data to AI 00:01:02 The DBT-Fivetran Merger: Not the End of the Modern Data Stack 00:05:26 Data Catalogs and What Went Wrong 00:08:16 Data Infrastructure at AI Labs: Surprising Insights 00:10:13 The Crazy Funding Environment of 2024-2025 00:17:18 World Models: Hype, Confusion, and Market Potential 00:18:59 Memory Management and Continual Learning: The Next Frontier 00:23:27 Agent Environments: Just a Fad? 00:25:48 The Perfect AI Startup: Research Meets Application 00:28:02 Closing Thoughts and Where to Find Sarah

One year ago, Anthropic launched the Model Context Protocol (MCP)—a simple, open standard to connect AI applications to the data and tools they need. Today, MCP has exploded from a local-only experiment into the de facto protocol for agentic systems, adopted by OpenAI, Microsoft, Google, Block, and hundreds of enterprises building internal agents at scale. And now, MCP is joining the newly formed Agentic AI Foundation (AAIF) under the Linux Foundation, alongside Block's Goose coding agent, with founding members spanning the biggest names in AI and cloud infrastructure. We sat down with David Soria Parra (MCP lead, Anthropic), Nick Cooper (OpenAI), Brad Howes (Block / Goose), and Jim Zemlin (Linux Foundation CEO) to dig into the one-year journey of MCP—from Thanksgiving hacking sessions and the first remote authentication spec to long-running tasks, MCP Apps, and the rise of agent-to-agent communication—and the behind-the-scenes story of how three competitive AI labs came together to donate their protocols and agents to a neutral foundation, why enterprises are deploying MCP servers faster than anyone expected (most of it invisible, internal, and at massive scale), what it takes to design a protocol that works for both simple tool calls and complex multi-agent orchestration, how the foundation will balance taste-making (curating meaningful projects) with openness (avoiding vendor lock-in), and the 2025 vision: MCP as the communication layer for asynchronous, long-running agents that work while you sleep, discover and install their own tools, and unlock the next order of magnitude in AI productivity. We discuss: The one-year MCP journey: from local stdio servers to remote HTTP streaming, OAuth 2.1 authentication (and the enterprise lessons learned), long-running tasks, and MCP Apps (iframes for richer UI) Why MCP adoption is exploding internally at enterprises: invisible, internal servers connecting agents to Slack, Linear, proprietary data, and compliance-heavy workflows (financial services, healthcare) The authentication evolution: separating resource servers from identity providers, dynamic client registration, and why the March spec wasn't enterprise-ready (and how June fixed it) How Anthropic dogfoods MCP: internal gateway, custom servers for Slack summaries and employee surveys, and why MCP was born from "how do I scale dev tooling faster than the company grows?" Tasks: the new primitive for long-running, asynchronous agent operations—why tools aren't enough, how tasks enable deep research and agent-to-agent handoffs, and the design choice to make tasks a "container" (not just async tools) MCP Apps: why iframes, how to handle styles and branding, seat selection and shopping UIs as the killer use case, and the collaboration with OpenAI to build a common standard The registry problem: official registry vs. curated sub-registries (Smithery, GitHub), trust levels, model-driven discovery, and why MCP needs "npm for agents" (but with signatures and HIPAA/financial compliance) The founding story of AAIF: how Anthropic, OpenAI, and Block came together (spoiler: they didn't know each other were talking to Linux Foundation), why neutrality matters, and how Jim Zemlin has never seen this much day-one inbound interest in 22 years — David Soria Parra (Anthropic / MCP) MCP: https://modelcontextprotocol.io https://uk.linkedin.com/in/david-soria-parra-4a78b3a https://x.com/dsp_ Nick Cooper (OpenAI) X: https://x.com/nicoaicopr Brad Howes (Block / Goose) Goose: https://github.com/block/goose Jim Zemlin (Linux Foundation) LinkedIn: https://www.linkedin.com/in/zemlin/ Agentic AI Foundation https://agenticai.foundation Chapters 00:00:00 Introduction: MCP's First Year and Foundation Launch 00:01:17 MCP's Journey: From Launch to Industry Standard 00:02:06 Protocol Evolution: Remote Servers and Authentication 00:08:52 Enterprise Authentication and Financial Services 00:11:42 Transport Layer Challenges: HTTP Streaming and Scalability 00:15:37 Standards Development: Collaboration with Tech Giants 00:34:27 Long-Running Tasks: The Future of Async Agents 00:30:41 Discovery and Registries: Building the MCP Ecosystem 00:30:54 MCP Apps and UI: Beyond Text Interfaces 00:26:55 Internal Adoption: How Anthropic Uses MCP 00:23:15 Skills vs MCP: Complementary Not Competing 00:36:16 Community Events and Enterprise Learnings 01:03:31 Foundation Formation: Why Now and Why Together 01:07:38 Linux Foundation Partnership: Structure and Governance 01:11:13 Goose as Reference Implementation 01:17:28 Principles Over Roadmaps: Composability and Quality 01:21:02 Foundation Value Proposition: Why Contribute 01:27:49 Practical Investments: Events, Tools, and Community 01:34:58 Looking Ahead: Async Agents and Real Impact

Note: Steve and Gene’s talk on Vibe Coding and the post IDE world was one of the top talks of AIE CODE: https://www.youtube.com/watch?v=7Dtu2bilcFs&t=1019s&pp=0gcJCU0KAYcqIYzv From building legendary platforms at Google and Amazon to authoring one of the most influential essays on AI-powered development (Revenge of the Junior Developer, quoted by Dario Amodei himself), Steve Yegge has spent decades at the frontier of software engineering—and now he's leading the charge into what he calls the "factory farming" era of code. After stints at SourceGraph and building Beads (a purely vibe-coded issue tracker with tens of thousands of users), Steve co-authored The Vibe Coding Book and is now building VC (VibeCoder), an agent orchestration dashboard designed to move developers from writing code to managing fleets of AI agents that coordinate, parallelize, and ship features while you sleep. We sat down with Steve at AI Engineer Summit to dig into why Claude Code, Cursor, and the entire 2024 stack are already obsolete, what it actually takes to trust an agent after 2,000 hours of practice (hint: they will delete your production database if you anthropomorphize them), why the real skill is no longer writing code but orchestrating agents like a NASCAR pit crew, how merging has become the new wall that every 10x-productive team is hitting (and why one company's solution is literally "one engineer per repo"), the rise of multi-agent workflows where agents reserve files, message each other via MCP, and coordinate like a little village, why Steve believes if you're still using an IDE to write code by January 1st, you're a bad engineer, how the 12–15 year experience bracket is the most resistant demographic (and why their identity is tied to obsolete workflows), the hidden chaos inside OpenAI, Anthropic, and Google as they scale at breakneck speed, why rewriting from scratch is now faster than refactoring for a growing class of codebases, and his 2025 prediction: we're moving from subsistence agriculture to John Deere-scale factory farming of code, and the Luddite backlash is only just beginning. We discuss: Why Claude Code, Cursor, and agentic coding tools are already last year's tech—and what comes next: agent orchestration dashboards where you manage fleets, not write lines The 2,000-hour rule: why it takes a full year of daily use before you can predict what an LLM will do, and why trust = predictability, not capability Steve's hot take: if you're still using an IDE to develop code by January 1st, 2025, you're a bad engineer—because the abstraction layer has moved from models to full-stack agents The demographic most resistant to vibe coding: 12–15 years of experience, senior engineers whose identity is tied to the way they work today, and why they're about to become the interns Why anthropomorphizing LLMs is the biggest mistake: the "hot hand" fallacy, agent amnesia, and how Steve's agent once locked him out of prod by changing his password to "fix" a problem Should kids learn to code? Steve's take: learn to vibe code—understand functions, classes, architecture, and capabilities in a language-neutral way, but skip the syntax The 2025 vision: "factory farming of code" where orchestrators run Cloud Code, scrub output, plan-implement-review-test in loops, and unlock programming for non-programmers at scale — Steve Yegge X: https://x.com/steve_yegge Substack (Stevie's Tech Talks): https://steve-yegge.medium.com/ GitHub (VC / VibeCoder): https://github.com/yegge-labs Where to find Latent Space X: https://x.com/latentspacepod Substack: https://www.latent.space/ Chapters 00:00:00 Introduction: Steve Yegge on Vibe Coding and AI Engineering 00:00:59 The Backlash: Who Resists Vibe Coding and Why 00:04:26 The 2000 Hour Rule: Building Trust with AI Coding Tools 00:03:31 The January 1st Deadline: IDEs Are Becoming Obsolete 00:02:55 10X Productivity at OpenAI: The Performance Review Problem 00:07:49 The Hot Hand Fallacy: When AI Agents Betray Your Trust 00:11:12 Claude Code Isn't It: The Need for Agent Orchestration 00:15:20 The Orchestrator Revolution: From Cloud Code to Agent Villages 00:18:46 The Merge Wall: The Biggest Unsolved Problem in AI Coding 00:26:33 Never Rewrite Your Code - Until Now: Joel Spolsky Was Wrong 00:22:43 Factory Farming Code: The John Deere Era of Software 00:29:27 Google's Gemini Turnaround and the AI Lab Chaos 00:33:20 Should Your Kids Learn to Code? The New Answer 00:34:59 Code MCP and the Gossip Rate: Latest Vibe Coding Discoveries

From the frontlines of OpenAI's Codex and GPT-5 training teams, Bryan and Bill are building the future of AI-powered coding—where agents don't just autocomplete, they architect, refactor, and ship entire features while you sleep. We caught up with them at AI Engineer Conference right after the launch of Codex Max, OpenAI's newest long-running coding agent designed to work for 24+ hours straight, manage its own context, and spawn sub-agents to parallelize work across your entire codebase. We sat down with Bryan and Bill to dig into what it actually takes to train a model that developers trust—why personality, communication, and planning matter as much as raw capability, how Codex is trained with strong opinions about tools (it loves rg over grep, seriously), why the abstraction layer is moving from models to full-stack agents you can plug into VS Code or Zed, how OpenAI partners co-develop tool integrations and discover unexpected model habits (like renaming tools to match Codex's internal training), the rise of applied evals that measure real-world impact instead of academic benchmarks, why multi-turn evals are the next frontier (and Bryan's "job interview eval" idea), how coding agents are breaking out of code into personal automation, terminal workflows, and computer use, and their 2026 vision: coding agents trusted enough to handle the hardest refactors at any company, not just top-tier firms, and general enough to build integrations, organize your desktop, and unlock capabilities you'd never get access to otherwise. We discuss: What Codex Max is: a long-running coding agent that can work 24+ hours, manage its own context window, and spawn sub-agents for parallel work Why the name "Max": maximalist, maximization, speed and endurance—it's simply better and faster for the same problems Training for personality: communication, planning, context gathering, and checking your work as behavioral characteristics, not just capabilities How Codex develops habits like preferring rg over grep, and why renaming tools to match its training (e.g., terminal-style naming) dramatically improves tool-call performance The split between Codex (opinionated, agent-focused, optimized for the Codex harness) and GPT-5 (general, more durable across different tools and modalities) Why the abstraction layer is moving up: from prompting models to plugging in full agents (Codex, GitHub Copilot, Zed) that package the entire stack The rise of sub-agents and agents-using-agents: Codex Max spawning its own instances, handing off context, and parallelizing work across a codebase How OpenAI works with coding partners on the bleeding edge to co-develop tool integrations and discover what the model is actually good at The shift to applied evals: capturing real-world use cases instead of academic benchmarks, and why ~50% of OpenAI employees now use Codex daily Why multi-turn evals are the next frontier: LM-as-a-judge for entire trajectories, Bryan's "job interview eval" concept, and the need for a batch multi-turn eval API How coding agents are breaking out of code: personal automation, organizing desktops, terminal workflows, and "Devin for non-coding" use cases Why Slack is the ultimate UI for work, and how coding agents can become your personal automation layer for email, files, and everything in between The 2026 vision: more computer use, more trust, and coding agents capable enough that any company can access top-tier developer capabilities, not just elite firms — Bryan & Bill (OpenAI Codex Team) http://x.com/bfioca https://x.com/realchillben OpenAI Codex: https://openai.com/index/openai-codex/ Where to find Latent Space X: https://x.com/latentspacepod Substack: https://www.latent.space/ Chapters 00:00:00 Introduction: Latent Space Listeners at AI Engineer Code 00:01:27 Codex Max Launch: Training for Long-Running Coding Agents 00:03:01 Model Personality and Trust: Communication, Planning, and Self-Checking 00:05:20 Codex vs GPT-5: Opinionated Agents vs General Models 00:07:47 Tool Use and Model Habits: The Ripgrep Discovery 00:09:16 Personality Design: Verbosity vs Efficiency in Coding Agents 00:11:56 The Agent Abstraction Layer: Building on Top of Codex 00:14:08 Sub-Agents and Multi-Agent Patterns: The Future of Composition 00:16:11 Trust and Adoption: OpenAI Developers Using Codex Daily 00:17:21 Applied Evals: Real-World Testing vs Academic Benchmarks 00:19:15 Multi-Turn Evals and the Job Interview Pattern 00:21:35 Feature Request: Batch Multi-Turn Eval API 00:22:28 Beyond Code: Personal Automation and Computer Use 00:24:51 Vision-Native Agents and the UI Integration Challenge 00:25:02 2026 Predictions: Trust, Computer Use, and Democratized Excellence

as with all demo-heavy and especially vision AI podcasts, we encourage watching along on our YouTube (and tossing us an upvote/subscribe if you like!) From SAM 1's 11-million-image data engine to SAM 2's memory-based video tracking, MSL’s Segment Anything project has redefined what's possible in computer vision. Now SAM 3 takes the next leap: concept segmentation—prompting with natural language like "yellow school bus" or "tablecloth" to detect, segment, and track every instance across images and video, in real time, with human-level exhaustivity. And with the latest SAM Audio (https://x.com/aiatmeta/status/2000980784425931067?s=46), SAM can now even segment audio output! We sat down with Nikhila Ravi (SAM lead at Meta) and Pengchuan Zhang (SAM 3 researcher) alongside Joseph Nelson (CEO, Roboflow) to unpack how SAM 3 unifies interactive segmentation, open-vocabulary detection, video tracking, and more into a single model that runs in 30ms on images and scales to real-time video on multi-GPU setups. We dig into the data engine that automated exhaustive annotation from two minutes per image down to 25 seconds using AI verifiers fine-tuned on Llama, the new SACO (Segment Anything with Concepts) benchmark with 200,000+ unique concepts vs. the previous 1.2k, how SAM 3 separates recognition from localization with a presence token, why decoupling the detector and tracker was critical to preserve object identity in video, how SAM 3 Agents unlock complex visual reasoning by pairing SAM 3 with multimodal LLMs like Gemini, and the real-world impact: 106 million smart polygons created on Roboflow saving humanity an estimated 130+ years of labeling time across fields from cancer research to underwater trash cleanup to autonomous vehicle perception. We discuss: What SAM 3 is: a unified model for concept-prompted segmentation, detection, and tracking in images and video using atomic visual concepts like "purple umbrella" or "watering can" How concept prompts work: short text phrases that find all instances of a category without manual clicks, plus visual exemplars (boxes, clicks) to refine and adapt on the fly Real-time performance: 30ms per image (100 detected objects on H200), 10 objects on 2×H200 video, 28 on 4×, 64 on 8×, with parallel inference and "fast mode" tracking The SACO benchmark: 200,000+ unique concepts vs. 1.2k in prior benchmarks, designed to capture the diversity of natural language and reach human-level exhaustivity The data engine: from 2 minutes per image (all-human) to 45 seconds (model-in-loop proposals) to 25 seconds (AI verifiers for mask quality and exhaustivity checks), fine-tuned on Llama 3.2 Why exhaustivity is central: every instance must be found, verified by AI annotators, and manually corrected only when the model misses—automating the hardest part of segmentation at scale Architecture innovations: presence token to separate recognition ("is it in the image?") from localization ("where is it?"), decoupled detector and tracker to preserve identity-agnostic detection vs. identity-preserving tracking Building on Meta's ecosystem: Perception Encoder, DINO v2 detector, Llama for data annotation, and SAM 2's memory-based tracking backbone SAM 3 Agents: using SAM 3 as a visual tool for multimodal LLMs (Gemini, Llama) to solve complex visual reasoning tasks like "find the bigger character" or "what distinguishes male from female in this image" Fine-tuning with as few as 10 examples: domain adaptation for specialized use cases (Waymo vehicles, medical imaging, OCR-heavy scenes) and the outsized impact of negative examples Real-world impact at Roboflow: 106M smart polygons created, saving 130+ years of labeling time across cancer research, underwater trash cleanup, autonomous drones, industrial automation, and more — MSL FAIR team Nikhila: https://www.linkedin.com/in/nikhilaravi/ Pengchuan: https://pzzhang.github.io/pzzhang/ Joseph Nelson X: https://x.com/josephofiowa LinkedIn: https://www.linkedin.com/in/josephofiowa/ [FLIGHTCAST_CHATPERS]

Note: this is Pliny and John’s first major podcast. Voices have been changed for opsec. From jailbreaking every frontier model and turning down Anthropic's Constitutional AI challenge to leading BT6, a 28-operator white-hat hacker collective obsessed with radical transparency and open-source AI security, Pliny the Liberator and John V are redefining what AI red-teaming looks like when you refuse to lobotomize models in the name of "safety." Pliny built his reputation crafting universal jailbreaks—skeleton keys that obliterate guardrails across modalities—and open-sourcing prompt templates like Libertas, predictive reasoning cascades, and the infamous "Pliny divider" that's now embedded so deep in model weights it shows up unbidden in WhatsApp messages. John V, coming from prompt engineering and computer vision, co-founded the Bossy Discord (40,000 members strong) and helps steer BT6's ethos: if you can't open-source the data, we're not interested. Together they've turned down enterprise gigs, pushed back on Anthropic's closed bounties, and insisted that real AI security happens at the system layer—not by bubble-wrapping latent space. We sat down with Pliny and John to dig into the mechanics of hard vs. soft jailbreaks, why multi-turn crescendo attacks were obvious to hackers years before academia "discovered" them, how segmented sub-agents let one jailbroken orchestrator weaponize Claude for real-world attacks (exactly as Pliny predicted 11 months before Anthropic's recent disclosure), why guardrails are security theater that punishes capability while doing nothing for real safety, the role of intuition and "bonding" with models to navigate latent space, how BT6 vets operators on skill and integrity, why they believe Mech Interp and open-source data are the path forward (not RLHF lobotomization), and their vision for a future where spatial intelligence, swarm robotics, and AGI alignment research happen in the open—bootstrapped, grassroots, and uncompromising. We discuss: What universal jailbreaks are: skeleton-key prompts that obliterate guardrails across models and modalities, and why they're central to Pliny's mission of "liberation" Hard vs. soft jailbreaks: single-input templates vs. multi-turn crescendo attacks, and why the latter were obvious to hackers long before academic papers The Libertas repo: predictive reasoning, the Library of Babel analogy, quotient dividers, weight-space seeds, and how introducing "steered chaos" pulls models out-of-distribution Why jailbreaking is 99% intuition and bonding with the model: probing token layers, syntax hacks, multilingual pivots, and forming a relationship to navigate latent space The Anthropic Constitutional AI challenge drama: UI bugs, judge failures, goalpost moving, the demand for open-source data, and why Pliny sat out the $30k bounty Why guardrails ≠ safety: security theater, the futility of locking down latent space when open-source is right behind, and why real safety work happens in meatspace (not RLHF) The weaponization of Claude: how segmented sub-agents let one jailbroken orchestrator execute malicious tasks (pyramid-builder analogy), and why Pliny predicted this exact TTP 11 months before Anthropic's disclosure BT6 hacker collective: 28 operators across two cohorts, vetted on skill and integrity, radical transparency, radical open-source, and the magic of moving the needle on AI security, swarm intelligence, blockchain, and robotics — Pliny the Liberator X: https://x.com/elder_plinius GitHub (Libertas): https://github.com/elder-plinius/L1B3RT45 John V X: https://x.com/JohnVersus BT6 & Bossy BT6: https://bt6.gg Bossy Discord: Search "Bossy Discord" or ask Pliny/John V on X Where to find Latent Space X: https://x.com/latentspacepod Substack: https://www.latent.space/ Chapters 00:00:00 Introduction: Meet Pliny the Liberator and John V 00:01:50 The Philosophy of AI Liberation and Jailbreaking 00:03:08 Universal Jailbreaks: Skeleton Keys to AI Models 00:04:24 The Cat-and-Mouse Game: Attackers vs Defenders 00:05:42 Security Theater vs Real Safety: The Fundamental Disconnect 00:08:51 Inside the Libertas Repo: Prompt Engineering as Art 00:16:22 The Anthropic Challenge Drama: UI Bugs and Open Source Data 00:23:30 From Jailbreaks to Weaponization: AI-Orchestrated Attacks 00:26:55 The BT6 Hacker Collective and BASI Community 00:34:46 AI Red Teaming: Full Stack Security Beyond the Model 00:38:06 Safety vs Security: Meat Space Solutions and Final Thoughts

Glean started as a Kleiner Perkins incubation and is now a $7B, $200m ARR Enterprise AI leader. Now KP has tapped its own podcaster to lead it’s next big swing. From building go-to-market the hard way in startups (and scaling Palo Alto Networks’ public cloud business) to joining Kleiner Perkins to help technical founders turn product edge into repeatable revenue, Joubin Mirzadegan has spent the last decade obsessing over one thing: distribution and how ideas actually spread, sell, and compound. That obsession took him from launching the CRO-only podcast Grit (https://www.youtube.com/playlist?list=PLRiWZFltuYPF8A6UGm74K2q29UwU-Kk9k) as a hiring wedge, to working alongside breakout companies like Glean and Windsurf, to now incubating Roadrunner which is an AI-native rethink of CPQ and quoting workflows as pricing models collapse from “seats” into consumption, bundles, renewals, and SKU sprawl. We sat down with Joubin to dig into the real mechanics of making conversations feel human (rolling early, never sending questions, temperature + lighting hacks), what Windsurf got right about “Google-class product and Salesforce-class distribution,” how to hire early sales leaders without getting fooled by shiny logos, why CPQ is quietly breaking the back of modern revenue teams, and his thesis for his new company and KP incubation Roadrunner (https://www.roadrunner.ai/): rebuild the data model from the ground up, co-develop with the hairiest design partners, and eventually use LLMs to recommend deal structures the way the best reps do without the Slack-channel chaos of deal desk. We discuss: How to make guests instantly comfortable: rolling early, no “are you ready?”, temperature, lighting, and room dynamics Why Joubin refuses to send questions in advance (and when you might have to anyway) The origin of the CRO-only podcast: using media as a hiring wedge and relationship engine The “commit to 100 episodes” mindset: why most shows die before they find their voice Founder vs exec interviews: why CEOs can speak more freely (and what it unlocks in conversation) What Glean taught him about enterprise AI: permissions, trust, and overcoming “category is dead” skepticism Design partners as the real unlock: why early believers matter and how co-development actually works Windsurf’s breakout: what it means to be serious about “Google-class product + Salesforce-class distribution” Why technical founders struggle with GTM and how KP built a team around sales, customer access, and demand gen Hiring early sales leaders: anti-patterns (logos), what to screen for (motivation), and why stage-fit is everything The CPQ problem & Roadrunner’s thesis: rebuilding CPQ/quoting from the data model up for modern complexity How “rules + SKUs + approvals” create a brittle graph and what it takes to model it without tipping over The two-year window: incumbents rebuilding slowly vs startups out-sprinting with AI-native architecture Where AI actually helps: quote generation, policy enforcement, approval routing, and deal recommendation loops — Joubin X: https://x.com/Joubinmir LinkedIn: https://www.linkedin.com/in/joubin-mirzadegan-66186854/ Where to find Latent Space X: https://x.com/latentspacepod Substack: https://www.latent.space/ Chapters 00:00:00 Introduction and the Zuck Interview Experience 00:03:26 The Genesis of the Grit Podcast: Hiring CROs Through Content 00:13:20 Podcast Philosophy: Creating Authentic Conversations 00:15:44 Working with Arvind at Glean: The Enterprise Search Breakthrough 00:26:20 Windsurf's Sales Machine: Google-Class Product Meets Salesforce-Class Distribution 00:30:28 Hiring Sales Leaders: Anti-Patterns and First Principles 00:39:02 The CPQ Problem: Why Salesforce and Legacy Tools Are Breaking 00:43:40 Introducing Roadrunner: Solving Enterprise Pricing with AI 00:49:19 Building Roadrunner: Team, Design Partners, and Data Model Challenges 00:59:35 High Performance Philosophy: Working Out Every Day and Reducing Friction 01:06:28 Defining Grit: Passion Plus Perseverance

From applied cryptography and offensive security in France’s defense industry to optimizing nuclear submarine workflows, then selling his e-signature startup to Docusign (https://www.docusign.com/company/news-center/opentrust-joins-docusign-global-trust-network and now running AI as CTO of Superhuman Mail (Superhuman, recently acquired by Grammarly https://techcrunch.com/2025/07/01/grammarly-acquires-ai-email-client-superhuman/), Loïc Houssier has lived the full arc from deep infra and compliance hell to obsessing over 100ms product experiences and AI-native email. We sat down with Loïc to dig into how you actually put AI into an inbox without adding latency, why Superhuman leans so hard into agentic search and “Ask AI” over your entire email history, how they design tools vs. agents and fight agent laziness, what box-priced inference and local-first caching mean for cost and reliability, and his bet that your inbox will power your future AI EA while AI massively widens the gap between engineers with real fundamentals and those faking it. We discuss: Loïc’s path from applied cryptography and offensive security in France’s defense industry to submarines, e-signatures, Docusign, and now Superhuman Mail What 3,000+ engineers actually do at a “simple” product like Docusign: regional compliance, on-prem appliances, and why global scale explodes complexity How Superhuman thinks about AI in email: auto-labels, smart summaries, follow-up nudges, “Ask AI” search, and the rule that AI must never add latency or friction Superhuman’s agentic framework: tools vs. agents, fighting “agent laziness,” deep semantic search over huge inboxes, and pagination strategies to find the real needle in the haystack How they evaluate OpenAI, Anthropic, Gemini, and open models: canonical queries, end-to-end evals, date reasoning, and Rahul’s infamous “what wood was my table?” test Infra and cost philosophy: local-first caching, vector search backends, Baseten “box” pricing vs. per-token pricing, and thinking in price-per-trillion-tokens instead of price-per-million The vision of Superhuman as your AI EA: auto-drafting replies in your voice, scheduling on your behalf, and using your inbox as the ultimate private data source How the Grammarly + Coda + Superhuman stack could power truly context-aware assistance across email, docs, calendars, contracts, and more Inside Superhuman’s AI-dev culture: free-for-all tool adoption, tracking AI usage on PRs, and going from ~4 to ~6 PRs per engineer per week Why Loïc believes everyone should still learn to code, and how AI will amplify great engineers with strong fundamentals while exposing shallow ones even faster — Loïc Houssier LinkedIn: https://www.linkedin.com/in/houssier/ Where to find Latent Space X: https://x.com/latentspacepod Substack: https://www.latent.space/ Chapters 00:00:00 Introduction and Loïc's Journey from Nuclear Submarines to Superhuman 00:06:40 Docusign Acquisition and the Enterprise Email Stack 00:10:26 Superhuman's AI Vision: Your Inbox as the Real AI Agent 00:13:20 Ask AI: Agentic Search and the Quality Problem 00:18:20 Infrastructure Choices: Model Selection, Base10, and Cost Management 00:27:30 Local-First Architecture and the Database Stack 00:30:50 Evals, Quality, and the Rahul Wood Table Test 00:42:30 The Future EA: Auto-Drafting and Proactive Assistance 00:46:40 Grammarly Acquisition and the Contextual Advantage 00:38:40 Voice, Video, and the End of Writing 00:51:40 Knowledge Graphs: The Hard Problem Nobody Has Solved 00:56:40 Competing with OpenAI and the Browser Question 01:02:30 AI Coding Tools: From 4 to 6 PRs Per Week 01:08:00 Engineering Culture, Hiring, and the Future of Software Development

From building Medal into a 12M-user game clipping platform with 3.8B highlight moments to turning down a reported $500M offer from OpenAI (https://www.theinformation.com/articles/openai-offered-pay-500-million-startup-videogame-data) and raising a $134M seed from Khosla (https://techcrunch.com/2025/10/16/general-intuition-lands-134m-seed-to-teach-agents-spatial-reasoning-using-video-game-clips/) to spin out General Intuition, Pim is betting that world models trained on peak human gameplay are the next frontier after LLMs. We sat down with Pim to dig into why game highlights are “episodic memory for simulation” (and how Medal’s privacy-first action labels became a world-model goldmine https://medal.tv/blog/posts/enabling-state-of-the-art-security-and-protections-on-medals-new-apm-and-controller-overlay-features), what it takes to build fully vision-based agents that just see frames and output actions in real time, how General Intuition transfers from games to real-world video and then into robotics, why world models and LLMs are complementary rather than rivals, what founders with proprietary datasets should know before selling or licensing to labs, and his bet that spatial-temporal foundation models will power 80% of future atoms-to-atoms interactions in both simulation and the real world. We discuss: How Medal’s 3.8B action-labeled highlight clips became a privacy-preserving goldmine for world models Building fully vision-based agents that only see frames and output actions yet play like (and sometimes better than) humans Transferring from arcade-style games to realistic games to real-world video using the same perception–action recipe Why world models need actions, memory, and partial observability (smoke, occlusion, camera shake) vs. “just” pretty video generation Distilling giant policies into tiny real-time models that still navigate, hide, and peek corners like real players Pim’s path from RuneScape private servers, Tourette’s, and reverse engineering to leading a frontier world-model lab How data-rich founders should think about valuing their datasets, negotiating with big labs, and deciding when to go independent GI’s first customers: replacing brittle behavior trees in games, engines, and controller-based robots with a “frames in, actions out” API Using Medal clips as “episodic memory of simulation” to move from imitation learning to RL via world models and negative events The 2030 vision: spatial–temporal foundation models that power the majority of atoms-to-atoms interactions in simulation and the real world — Pim X: https://x.com/PimDeWitte LinkedIn: https://www.linkedin.com/in/pimdw/ Where to find Latent Space X: https://x.com/latentspacepod Substack: https://www.latent.space/ Chapters 00:00:00 Introduction and Medal's Gaming Data Advantage 00:02:08 Exclusive Demo: Vision-Based Gaming Agents 00:06:17 Action Prediction and Real-World Video Transfer 00:08:41 World Models: Interactive Video Generation 00:13:42 From Runescape to AI: Pim's Founder Journey 00:16:45 The Research Foundations: Diamond, Genie, and SEMA 00:33:03 Vinod Khosla's Largest Seed Bet Since OpenAI 00:35:04 Data Moats and Why GI Stayed Independent 00:38:42 Self-Teaching AI Fundamentals: The Francois Fleuret Course 00:40:28 Defining World Models vs Video Generation 00:41:52 Why Simulation Complexity Favors World Models 00:43:30 World Labs, Yann LeCun, and the Spatial Intelligence Race 00:50:08 Business Model: APIs, Agents, and Game Developer Partnerships 00:58:57 From Imitation Learning to RL: Making Clips Playable 01:00:15 Open Research, Academic Partnerships, and Hiring 01:02:09 2030 Vision: 80 Percent of Atoms-to-Atoms AI Interactions

Fei-Fei Li and Justin Johnson are cofounders of World Labs, who have recently launched Marble (https://marble.worldlabs.ai/), a new kind of generative “world model” that can create editable 3D environments from text, images, and other spatial inputs. Marble lets creators generate persistent 3D worlds, precisely control cameras, and interactively edit scenes, making it a powerful tool for games, film, VR, robotics simulation, and more. In this episode, Fei-Fei and Justin share how their journey from ImageNet and Stanford research led to World Labs, why spatial intelligence is the next frontier after LLMs, and how world models could change how machines see, understand, and build in 3D. We discuss: The massive compute scaling from AlexNet to today and why world models and spatial data are the most compelling way to “soak up” modern GPU clusters compared to language alone. What Marble actually is: a generative model of 3D worlds that turns text and images into editable scenes using Gaussian splats, supports precise camera control and recording, and runs interactively on phones, laptops, and VR headsets. Fei-fei’s essay (https://drfeifei.substack.com/p/from-words-to-worlds-spatial-intelligence) on spatial intelligence as a distinct form of intelligence from language: from picking up a mug to inferring the 3D structure of DNA, and why language is a lossy, low-bandwidth channel for describing the rich 3D/4D world we live in. Whether current models “understand” physics or just fit patterns: the gap between predicting orbits and discovering F=ma, and how attaching physical properties to splats and distilling physics engines into neural networks could lead to genuine causal reasoning. The changing role of academia in AI, why Fei-Fei worries more about under-resourced universities than “open vs closed,” and how initiatives like national AI compute clouds and open benchmarks can rebalance the ecosystem. Why transformers are fundamentally set models, not sequence models, and how that perspective opens up new architectures for world models, especially as hardware shifts from single GPUs to massive distributed clusters. Real use cases for Marble today: previsualization and VFX, game environments, virtual production, interior and architectural design (including kitchen remodels), and generating synthetic simulation worlds for training embodied agents and robots. How spatial intelligence and language intelligence will work together in multimodal systems, and why the goal isn’t to throw away LLMs but to complement them with rich, embodied models of the world. Fei-Fei and Justin’s long-term vision for spatial intelligence: from creative tools for artists and game devs to broader applications in science, medicine, and real-world decision-making. — Fei-Fei Li X: https://x.com/drfeifei LinkedIn: https://www.linkedin.com/in/fei-fei-li-4541247 Justin Johnson X: https://x.com/jcjohnss LinkedIn: https://www.linkedin.com/in/justin-johnson-41b43664 Where to find Latent Space X: https://x.com/latentspacepod Substack: https://www.latent.space/ Chapters 00:00:00 Introduction and the Fei-Fei Li & Justin Johnson Partnership 00:02:00 From ImageNet to World Models: The Evolution of Computer Vision 00:12:42 Dense Captioning and Early Vision-Language Work 00:19:57 Spatial Intelligence: Beyond Language Models 00:28:46 Introducing Marble: World Labs' First Spatial Intelligence Model 00:33:21 Gaussian Splats and the Technical Architecture of Marble 00:22:10 Physics, Dynamics, and the Future of World Models 00:41:09 Multimodality and the Interplay of Language and Space 00:37:37 Use Cases: From Creative Industries to Robotics and Embodied AI 00:56:58 Hiring, Research Directions, and the Future of World Labs

Alex Lieberman and Arman Hezarkani, co-founders of Tenex, reveal how they're revolutionizing software consulting by compensating AI engineers for output rather than hours—enabling some engineers to earn over $1 million annually while delivering 10x productivity gains. Their company represents a fundamental rethinking of knowledge work compensation in the age of AI agents, where traditional hourly billing models perversely incentivize slower work even as AI tools enable unprecedented speed. The Genesis: From 90% Downsizing to 10x Output The story behind 10X begins with Arman's previous company, Parthian, where he was forced to downsize his engineering team by 90%. Rather than collapse, Arman re-architected the entire product and engineering process to be AI-first—and discovered that production-ready software output increased 10x despite the massive headcount reduction. This counterintuitive result exposed a fundamental misalignment: engineers compensated by the hour are disincentivized from leveraging AI to work faster, even when the technology enables dramatic productivity gains. Alex, who had invested in Parthian, initially didn't believe the numbers until Arman walked him through why LLMs have made such a profound impact specifically on engineering as knowledge work. The Economic Model: Story Points Over Hours 10X's core innovation is compensating engineers based on story points—units of completed, quality output—rather than hours worked. This creates direct economic incentives for engineers to adopt every new AI tool, optimize their workflows, and maximize throughput. The company expects multiple engineers to earn over $1 million in cash compensation next year purely from story point earnings. To prevent gaming the system, they hire for two profiles: engineers who are "long-term selfish" (understanding that inflating story points will destroy client relationships) and those who genuinely love writing code and working with smart people. They also employ technical strategists incentivized on client retention (NRR) who serve as the final quality gate before any engineering plan reaches a client. Impressive Builds: From Retail AI to App Store Hits The results speak for themselves. In one project, 10X built a computer vision system for retail cameras that provides heat maps, queue detection, shelf stocking analysis, and theft detection—creating early prototypes in just two weeks for work that previously took quarters. They built Snapback Sports' mobile trivia app in one month, which hit 20th globally on the App Store. In a sales context, an engineer spent four hours building a working prototype of a fitness influencer's AI health coach app after the prospect initially said no—immediately moving 10X to the top of their vendor list. These examples demonstrate how AI-enabled speed fundamentally changes sales motions and product development timelines. The Interview Process: Unreasonably Difficult Take-Homes Despite concerns that AI would make take-home assessments obsolete, 10X still uses them—but makes them "unreasonably difficult." About 50% of candidates don't even respond, but those who complete the challenge demonstrate the caliber needed. The interview process is remarkably short: two calls before the take-home, review, then one or two final meetings—completable in as little as a week. A signature question: "If you had infinite resources to build an AI that could replace either of us on this call, what would be the first major bottleneck?" The sophisticated answer isn't just "model intelligence" or "context length"—it's controlling entropy, the accumulating error rate that derails autonomous agents over time. The Limiting Factor: Human Capital, Not Technology Despite being an AI-first company, 10X's primary constraint is human capital—finding and hiring enough exceptional engineers fast enough, then matching them with the right processes to maintain delivery quality as they scale. The company has ambitions beyond consulting to build their own technology, but for the foreseeable future, recruiting remains the bottleneck. This reveals an important insight about the AI era: even as technology enables unprecedented leverage, the constraint shifts to finding people who can harness that leverage effectively. Chapters 00:00:00 Introduction and Meeting the 10X Co-founders 00:01:29 The 10X Moment: From Hourly Billing to Output-Based Compensation 00:04:44 The Economic Model Behind 10X 00:05:42 Story Points and Measuring Engineering Output 00:08:41 Impressive Client Projects and Rapid Prototyping 00:12:22 The 10X Tech Stack: TypeScript and High Structure 00:13:21 AI Coding Tools: The Daily Evolution 00:15:05 Human Capital as the Limiting Factor 00:16:02 The Unreasonably Difficult Interview Process 00:17:14 Entropy and Context Engineering: The Future of AI Agents 00:23:28 The MCP Debate and AI Industry Sociology 00:26:01 Consulting, Digital Transformation, and Conference Insights

Deedy Das, Partner at Menlo Ventures, returns to Latent Space to discuss his journey from Glean to venture capital, the explosive rise of Anthropic, and how AI is reshaping enterprise software and coding. From investing in Anthropic early on when they had no revenue to managing the $100M Ontology Fund, Das shares insider perspectives on the fastest-growing software company in history and what's next for AI infrastructure, research investing, and the future of engineering. We cover Glean’s rise from “boring” enterprise search to a $7B AI-native company, Anthropic's meteoric rise, the strategic decisions behind products like Claude Code, and why market share in enterprise AI is shifting dramatically. Das explains his investment thesis on research companies like Goodfire, Prime Intellect, and OpenRouter and how the Anthology Fund is quietly seeding the next wave of AI infra, research, and devtools. Chapters 00:00:00 Introduction and Deedy's Return to Latent Space 00:01:20 Glean's Journey: From Boring Enterprise Search to $7B Valuation 00:15:37 Anthropic's Meteoric Rise and Market Share Dynamics 00:17:50 Claude Artifacts and Product Innovation 00:41:20 The Anthology Fund: Investing in the Anthropic Ecosystem 00:48:01 Goodfire and Mechanistic Interpretability 00:51:25 Prime Intellect and Distributed AI Training 00:53:40 OpenRouter: Building the AI Model Gateway 01:13:36 The Stargate Project and Infrastructure Arms Race 01:18:14 The Future of Software Engineering and AI Coding

Jared Palmer, SVP at GitHub and VP of CoreAI at Microsoft, joins Latent Space for an in-depth look at the evolution of coding agents and modern developer tools. Recently joining after leading AI initiatives at Vercel, Palmer shares firsthand insights from behind the scenes at GitHub Universe, including the launch of Agent HQ which is a new collaboration hub for coding agents and developers. This episode traces Palmer’s journey from building Copilot inspired tools to pioneering the focused Next.js coding agent, v0, and explores how platform constraints fostered rapid experimentation and a breakout success in AI-powered frontend development. Palmer explains the unique advantages of GitHub’s massive developer network, the challenges of scaling agent-based workflows, and why integrating seamless AI into developer experiences is now a top priority for both Microsoft and GitHub.

Jed Borovik, Product Lead at Google Labs, joins Latent Space to unpack how Google is building the future of AI-powered software development with Jules. From his journey discovering GenAI through Stable Diffusion to leading one of the most ambitious coding agent projects in tech, Borovik shares behind-the-scenes insights into how Google Labs operates at the intersection of DeepMind's model development and product innovation. We explore Jules' approach to autonomous coding agents and why they run on their own infrastructure, how Google simplified their agent scaffolding as models improved, and why embeddings-based RAG is giving way to attention-based search. Borovik reveals how developers are using Jules for hours or even days at a time, the challenges of managing context windows that push 2 million tokens, and why coding agents represent both the most important AI application and the clearest path to AGI. This conversation reveals Google's positioning in the coding agent race, the evolution from internal tools to public products, and what founders, developers, and AI engineers should understand about building for a future where AI becomes the new brush for software engineering. Chapters 00:00:00 Introduction and GitHub Universe Recap 00:00:57 New York Tech Scene and East Coast Hackathons 00:02:19 From Google Search to AI Coding: Jed's Journey 00:04:19 Google Labs Mission and DeepMind Collaboration 00:06:41 Jules: Autonomous Coding Agents Explained 00:09:39 The Evolution of Agent Scaffolding and Model Quality 00:11:30 RAG vs Attention: The Shift in Code Understanding 00:13:49 Jules' Journey from Preview to Production 00:15:05 AI Engineer Summit: Community Building and Networking 00:25:06 Context Management in Long-Running Agents 00:29:02 The Future of Software Engineering with AI 00:36:26 Beyond Vibe Coding: Spec Development and Verification 00:40:20 Multimodal Input and Computer Use for Coding Agents

Today’s guests are Priscilla Chan and Mark Zuckerberg, co-founders of Biohub (fka Chan Zuckerberg Initiative). They are one of the leading institutes for AI x Bio and open science research with projects like CELLxGENE, rbio1, VariantFormer, and many more. We talked about the evolution from a broad philanthropic institute to specializing in frontier AI + bio, why they are building 12ft tall microscopes to gather better data, and how building a virtual cell model + virtual immune system could potentially help us cure all diseases. Chapters 00:00:00 Introduction and CZI's 10-Year Anniversary 00:00:56 Learning from Bill Gates 00:04:05 Science vs Translation 00:10:45 The Power of Physical Proximity in Science 00:13:55 Building the Virtual Cell: From Data to Models 00:15:51 Microscopes, Imaging, and Converting Atoms to Bits 00:23:18 AI Meets Biology: The Frontier Lab Concept 00:27:25 How Models Can Enable More Ambitious Research 00:30:15 Precision Medicine and Clinical Impact 00:45:17 The Virtual Immune System and Cellular Engineering 00:48:27 Accelerating the Timeline: What It Takes to Cure All Disease 00:28:45 Joining Forces with Evolutionary Scale

In this conversation with Malte Ubl, CTO of Vercel (http://x.com/cramforce), we explore how the company is pioneering the infrastructure for AI-powered development through their comprehensive suite of tools including workflows, AI SDK, and the newly announced agent ecosystem. Malte shares insights into Vercel's philosophy of "dogfooding" - never shipping abstractions they haven't battle-tested themselves - which led to extracting their AI SDK from v0 and building production agents that handle everything from anomaly detection to lead qualification. The discussion dives deep into Vercel's new Workflow Development Kit, which brings durable execution patterns to serverless functions, allowing developers to write code that can pause, resume, and wait indefinitely without cost. Malte explains how this enables complex agent orchestration with human-in-the-loop approvals through simple webhook patterns, making it dramatically easier to build reliable AI applications. We explore Vercel's strategic approach to AI agents, including their DevOps agent that automatically investigates production anomalies by querying observability data and analyzing logs - solving the recall-precision problem that plagues traditional alerting systems. Malte candidly discusses where agents excel today (meeting notes, UI changes, lead qualification) versus where they fall short, emphasizing the importance of finding the "sweet spot" by asking employees what they hate most about their jobs. The conversation also covers Vercel's significant investment in Python support, bringing zero-config deployment to Flask and FastAPI applications, and their vision for security in an AI-coded world where developers "cannot be trusted." Malte shares his perspective on how CTOs must transform their companies for the AI era while staying true to their core competencies, and why maintaining strong IC (individual contributor) career paths is crucial as AI changes the nature of software development. What was launched at Ship AI 2025: AI SDK 6.0 & Agent Architecture Agent Abstraction Philosophy: AI SDK 6 introduces an agent abstraction where you can "define once, deploy everywhere". How does this differ from existing agent frameworks like LangChain or AutoGPT? What specific pain points did you observe in production that led to this design? Human-in-the-Loop at Scale: The tool approval system with needsApproval: true gates actions until human confirmation. How do you envision this working at scale for companies with thousands of agent executions? What's the queue management and escalation strategy? Type Safety Across Models: AI SDK 6 promises "end-to-end type safety across models and UI". Given that different LLMs have varying capabilities and output formats, how do you maintain type guarantees when swapping between providers like OpenAI, Anthropic, or Mistral? Workflow Development Kit (WDK) Durability as Code: The use workflow primitive makes any TypeScript function durable with automatic retries, progress persistence, and observability. What's happening under the hood? Are you using event sourcing, checkpoint/restart, or a different pattern? Infrastructure Provisioning: Vercel automatically detects when a function is durable and dynamically provisions infrastructure in real-time. What signals are you detecting in the code, and how do you determine the optimal infrastructure configuration (queue sizes, retry policies, timeout values)? Vercel Agent (beta) Code Review Validation: The Agent reviews code and proposes "validated patches". What does "validated" mean in this context? Are you running automated tests, static analysis, or something more sophisticated? AI Investigations: Vercel Agent automatically opens AI investigations when it detects performance or error spikes using real production data. What data sources does it have access to? How does it distinguish between normal variance and actual anomalies? Python Support (For the first time, Vercel now supports Python backends natively.) Marketplace & Agent Ecosystem Agent Network Effects: The Marketplace now offers agents like CodeRabbit, Corridor, Sourcery, and integrations with Autonoma, Braintrust, Browser Use. How do you ensure these third-party agents can't access sensitive customer data? What's the security model? "An Agent on Every Desk" Program Vercel launched a new program to help companies identify high-value use cases and build their first production AI agents. It provides consultations, reference templates, and hands-on support to go from idea to deployed agent

Emily Glassberg Sands is the Head of Data & AI at Stripe where she leads the organization’s efforts to build financial infrastructure for the internet & leverage AI to power Stripe’s products. Stripe processes about $1.4 trillion in payments annually (~1.3% of global GDP), making it an exciting opportunity to apply AI & ML at scale. In this episode, Emily shares insights into how Stripe is using AI to solve complex problems like fraud detection, optimizing checkout experiences, & enabling new business models for AI companies. Emily also shares her economist perspective on market efficiency & how Stripe’s focus on building economic infrastructure for AI is driving growth across the ecosystem. We discuss: Stripe’s domain-specific foundation model and “payments embeddings” that run inline on the charge path to detect sophisticated card-testing at scale (improved detection rates at large users from ~59% to ~97%). The launch of the Agentic Commerce Protocol (ACP) with OpenAI, creating a shared standard for how businesses can expose products to AI agents which is used by Walmart and Sam’s Club. How Stripe is helping AI companies manage new fraud vectors, such as free trial and refund abuse, and the importance of real-time, outcome-based billing The impact of AI on Stripe’s internal operations, including the use of LLMs for code generation, merchant understanding, and internal tooling Why many AI companies are going global day-one how Stripe’s Link network (200M+ consumers) concentrates AI demand. Whether we're in an AI bubble, why GDP hasn't reflected AI productivity gains yet, and how agentic commerce could expand consumption by removing time constraints for high-income consumers Emily’s perspective on the changing social contract around AI, the importance of deep thinking, and the role of brand and design in AI-driven products — Where to find Emily Sands X: https://x.com/emilygsands LinkedIn: https://www.linkedin.com/in/egsands/ Where to find Shawn Wang X: https://x.com/swyx LinkedIn: https://www.linkedin.com/in/shawnswyxwang/ Where to find Alessio Fanelli X: https://x.com/FanaHOVA LinkedIn: https://www.linkedin.com/in/fanahova/ Where to find Latent Space X: https://x.com/latentspacepod Substack: https://www.latent.space/ Chapters 00:00:00 Introduction and Emily's Role at Stripe 00:09:55 AI Business Models and Fraud Challenges 00:13:49 Extending Radar for AI Economy 00:16:42 Payment Innovation: Token Billing and Stablecoins 00:23:09 Agentic Commerce Protocol Launch 00:29:40 Good Bots vs Bad Bots in AI 00:40:31 Designing the Agents Commerce Protocol 00:49:32 Internal AI Adoption at Stripe 01:04:53 Data Discovery and Text-to-SQL Challenges 01:21:00 AI Economy Analysis: Bubble or Boom?

In this deep dive with Kyle Corbitt, co-founder and CEO of OpenPipe (recently acquired by CoreWeave), we explore the evolution of fine-tuning in the age of AI agents and the critical shift from supervised fine-tuning to reinforcement learning. Kyle shares his journey from leading YC's Startup School to building OpenPipe, initially focused on distilling expensive GPT-4 workflows into smaller, cheaper models before pivoting to RL-based agent training as frontier model prices plummeted. The conversation reveals why 90% of AI projects remain stuck in proof-of-concept purgatory - not due to capability limitations, but reliability issues that Kyle believes can be solved through continuous learning from real-world experience. He discusses the breakthrough of RULER (Relative Universal Reinforcement Learning Elicited Rewards), which uses LLMs as judges to rank agent behaviors relatively rather than absolutely, making RL training accessible without complex reward engineering. Kyle candidly assesses the challenges of building realistic training environments for agents, explaining why GRPO (despite its advantages) may be a dead end due to its requirement for perfectly reproducible parallel rollouts. He shares insights on why LoRAs remain underrated for production deployments, why GEPA and prompt optimization haven't lived up to the hype in his testing, and why the hardest part of deploying agents isn't the AI - it's sandboxing real-world systems with all their bugs and edge cases intact. The discussion also covers OpenPipe's acquisition by CoreWeave, the launch of their serverless reinforcement learning platform, and Kyle's vision for a future where every deployed agent continuously learns from production experience. He predicts that solving the reliability problem through continuous RL could unlock 10x more AI inference demand from projects currently stuck in development, fundamentally changing how we think about agent deployment and maintenance. Key Topics: The rise and fall of fine-tuning as a business model Why 90% of AI projects never reach production RULER: Making RL accessible through relative ranking The environment problem: Why sandboxing is harder than training GRPO vs PPO and the future of RL algorithms LoRAs: The underrated deployment optimization Why GEPA and prompt optimization disappointed in practice Building world models as synthetic training environments The $500B Stargate bet and OpenAI's potential crypto play Continuous learning as the path to reliable agents References https://www.linkedin.com/in/kcorbitt/ Aug 2023  https://openpipe.ai/blog/from-prompts-to-models  DEC 2023 https://openpipe.ai/blog/mistral-7b-fine-tune-optimized JAN 2024 https://openpipe.ai/blog/s-lora MAY 2024 https://openpipe.ai/blog/the-ten-commandments-of-fine-tuning-in-prod   https://www.youtube.com/watch?v=-hYqt8M9u_M Oct 2024 https://openpipe.ai/blog/announcing-dpo-support  AIE NYC 2025 Finetuning 500m agents https://www.youtube.com/watch?v=zM9RYqCcioM&t=919s AIEWF 2025 How to train your agent (ART-E) https://www.youtube.com/watch?v=gEDl9C8s_-4&t=216s SEPT 2025 ACQUISTION https://openpipe.ai/blog/openpipe-coreweave  W&B Serverless RL https://openpipe.ai/blog/serverless-rl?refresh=1760042248153

At OpenAI DevDay, we sit down with Sherwin Wu and Christina Cai from the OpenAI Platform Team to discuss the launch of AgentKit - a comprehensive suite of tools for building, deploying, and optimizing AI agents. Christina walks us through the live demo she performed on stage, building a customer support agent in just 8 minutes using the visual Agent Builder, while Sherwin shares insights on how OpenAI is inverting the traditional website-chatbot paradigm by embedding apps directly within ChatGPT through the new Apps SDK. The conversation explores how OpenAI is tackling the challenges developers face when taking agents to production - from writing and optimizing prompts to building evaluation pipelines. They discuss the decision to adopt Anthropic's MCP protocol for tool connectivity, the importance of visual workflows for complex agent systems, and how features like human-in-the-loop approvals and automated prompt optimization are making agent development more accessible to a broader range of developers. Sherwin and Christina also reveal how OpenAI is dogfooding these tools internally, with their own customer support at openai.com already powered by AgentKit, and share candid insights about the evolution from plugins to GPTs to this new agent platform. They discuss the surprising persistence of prompting as a critical skill (contrary to predictions from two years ago), the challenges of serving custom fine-tuned models at scale, and why they believe visual agent builders are essential as workflows grow to span dozens of nodes. Guests: Sherwin Wu: Head of Engineering, OpenAI Platform https://www.linkedin.com/in/sherwinwu1/ https://x.com/sherwinwu?lang=en Christina Huang: Platform Experience, OpenAI https://x.com/christinaahuang https://www.linkedin.com/in/christinaahuang/ Thanks very much to Lindsay and Shaokyi for helping us set up this great deepdive into the new DevDay launches! Key Topics: • AgentKit launch: Agent SDK, Builder, Evals, and deployment tools • Apps SDK and the inversion of the app-chatbot paradigm • Adopting MCP protocol for universal tool connectivity • Visual agent building vs code-first approaches • Human-in-the-loop workflows and approval systems • Automated prompt optimization and "zero-gradient fine-tuning" • Service Health Dashboard and achieving five nines reliability • ChatKit as an embeddable, evergreen chat interface • The evolution from plugins to GPTs to agent platforms • Internal dogfooding with Codex and agent-powered support

Dylan Field (CEO Figma) on how they are letting designers build with Figma Make, how Figma can be the context repository for aesthetic in the age of vibe coding, and why design is your only differentiator now. Full show notes: https://www.latent.space/p/figma 00:00 Figma’s Mission: Bridging Imagination and Reality 00:56 Becoming AI-Pilled 07:44 Figma Make 08:57 Language as the Interface for Design 13:37 Source of truth between design and code 18:15 Figma as a Context Repository 21:30 Understanding and Representing Design Diffs through AI 24:20 Figma’s Role in Shaping Visual Aesthetics 31:56 Fast Fashion in Software 36:04 Limitations of Prompt-Based Software Creation 39:43 Interfaces Beyond Chat 42:12 Lessons from the Thiel Fellowship 44:58 Using X for Product Feedback 48:10 Early-Stage Recruiting at Figma 53:11 Positioning Figma Make in the Prompt-to-App Landscape 55:19 Digital Scarcity & AI

Quinn Slack (CEO) and Thorsten Ball (Amp Dictator) from SourceGraph join the show to talk about Amp Code, how they ship 15x/day with no code reviews, and why subagents and prompt optimizers aren’t a promising direction for coding agents. Amp Code: https://ampcode.com/ Latent Space: https://latent.space/ 00:00 Introduction 00:41 Transition from Cody to Amp 03:18 The Importance of Building the Best Coding Agent 06:43 Adapting to a Rapidly Evolving AI Tooling Landscape 09:36 Dogfooding at Sourcegraph 12:35 CLI vs. VS Code Extension 21:08 Positioning Amp in Coding Agent Market 24:10 The Diminishing Importance of Model Selectors 32:39 Tooling vs. Harness 37:19 Common Failure Modes of Coding Agents 47:33 Agent-Friendly Logging and Tooling 52:31 Are Subagents Real? 56:52 New Frameworks and Agent-Integrated Developer Tools 1:00:25 How Agents Are Encouraging Codebase and Workflow Changes 1:03:13 Evolving Outer Loop Tasks 1:07:09 Version Control and Merge Conflicts in an AI-First World 1:10:36 Rise of User-Generated Enterprise Software 1:14:39 Empowering Technical Leaders with AI 1:17:11 Evaluating Product Without Traditional Evals 1:20:58 Hiring

Lance: https://www.linkedin.com/in/lance-martin-64a33b5/ How Context Fails: https://www.dbreunig.com/2025/06/22/how-contexts-fail-and-how-to-fix-them.html How New Buzzwords Get Created: https://www.dbreunig.com/2025/07/24/why-the-term-context-engineering-matters.html Content Engineering: https://x.com/RLanceMartin/status/1948441848978309358 https://rlancemartin.github.io/2025/06/23/context_engineering/ https://docs.google.com/presentation/d/16aaXLu40GugY-kOpqDU4e-S0hD1FmHcNyF0rRRnb1OU/edit?usp=sharing Manus Post: https://manus.im/blog/Context-Engineering-for-AI-Agents-Lessons-from-Building-Manus Cognition Post: https://cognition.ai/blog/dont-build-multi-agents Multi-Agent Researcher: https://www.anthropic.com/engineering/multi-agent-research-system Human-in-the-loop + Memory: https://github.com/langchain-ai/agents-from-scratch - Bitter Lesson in AI Engineering - Hyung Won Chung on the Bitter Lesson in AI Research: https://www.youtube.com/watch?v=orDKvo8h71o Bitter Lesson w/ Claude Code: https://www.youtube.com/watch?v=Lue8K2jqfKk&t=1s Learning the Bitter Lesson in AI Engineering: https://rlancemartin.github.io/2025/07/30/bitter_lesson/ Open Deep Research: https://github.com/langchain-ai/open_deep_research https://academy.langchain.com/courses/deep-research-with-langgraph Scaling and building things that "don't yet work": https://www.youtube.com/watch?v=p8Jx4qvDoSo - Frameworks - Roast framework at Shopify / standardization of orchestration tools: https://www.youtube.com/watch?v=0NHCyq8bBcM MCP adoption within Anthropic / standardization of protocols: https://www.youtube.com/watch?v=xlEQ6Y3WNNI How to think about frameworks: https://blog.langchain.com/how-to-think-about-agent-frameworks/ RAG benchmarking: https://rlancemartin.github.io/2025/04/03/vibe-code/ Simon's talk with memory-gone-wrong: https://simonwillison.net/2025/Jun/6/six-months-in-llms/

Today we are joined by Gorkem and Batuhan from Fal.ai, the fastest growing generative media inference provider. They recently raised a $125M Series C and crossed $100M ARR. We covered how they pivoted from dbt pipelines to diffusion models inference, what were the models that really changed the trajectory of image generation, and the future of AI videos. Enjoy! 00:00 - Introductions 04:58 - History of Major AI Models and Their Impact on Fal.ai 07:06 - Pivoting to Generative Media and Strategic Business Decisions 10:46 - Technical discussion on CUDA optimization and kernel development 12:42 - Inference Engine Architecture and Kernel Reusability 14:59 - Performance Gains and Latency Trade-offs 15:50 - Discussion of model latency importance and performance optimization 17:56 - Importance of Latency and User Engagement 18:46 - Impact of Open Source Model Releases and Competitive Advantage 19:00 - Partnerships with closed source model developers 20:06 - Collaborations with Closed-Source Model Providers 21:28 - Serving Audio Models and Infrastructure Scalability 22:29 - Serverless GPU infrastructure and technical stack 23:52 - GPU Prioritization: H100s and Blackwell Optimization 25:00 - Discussion on ASICs vs. General Purpose GPUs 26:10 - Architectural Trends: MMDiTs and Model Innovation 27:35 - Rise and Decline of Distillation and Consistency Models 28:15 - Draft Mode and Streaming in Image Generation Workflows 29:46 - Generative Video Models and the Role of Latency 30:14 - Auto-Regressive Image Models and Industry Reactions 31:35 - Discussion of OpenAI's Sora and competition in video generation 34:44 - World Models and Creative Applications in Games and Movies 35:27 - Video Models’ Revenue Share and Open-Source Contributions 36:40 - Rise of Chinese Labs and Partnerships 38:03 - Top Trending Models on Hugging Face and ByteDance's Role 39:29 - Monetization Strategies for Open Models 40:48 - Usage Distribution and Model Turnover on FAL 42:11 - Revenue Share vs. Open Model Usage Optimization 42:47 - Moderation and NSFW Content on the Platform 44:03 - Advertising as a key use case for generative media 45:37 - Generative Video in Startup Marketing and Virality 46:56 - LoRA Usage and Fine-Tuning Popularity 47:17 - LoRA ecosystem and fine-tuning discussion 49:25 - Post-Training of Video Models and Future of Fine-Tuning 50:21 - ComfyUI Pipelines and Workflow Complexity 52:31 - Requests for startups and future opportunities in the space 53:33 - Data Collection and RedPajama-Style Initiatives for Media Models 53:46 - RL for Image and Video Models: Unknown Potential 55:11 - Requests for Models: Editing and Conversational Video Models 57:12 - VO3 Capabilities: Lip Sync, TTS, and Timing 58:23 - Bitter Lesson and the Future of Model Workflows 58:44 - FAL's hiring approach and team structure 59:29 - Team Structure and Scaling Applied ML and Performance Teams 1:01:41 - Developer Experience Tools and Low-Code/No-Code Integration 1:03:04 - Improving Hiring Process with Public Challenges and Benchmarks 1:04:02 - Closing Remarks and Culture at FAL

Our chat with Ari shows that data curation is the most impactful and underinvested area in AI. He argues that the prevailing focus on model architecture and compute scaling overlooks the "bitter lesson" that "models are what they eat." Effective data curation—a sophisticated process involving filtering, rebalancing, sequencing (curriculum), and synthetic data generation—allows for training models that are simultaneously faster, better, and smaller. Morcos recounts his personal journey from focusing on model-centric inductive biases to realizing that data quality is the primary lever for breaking the diminishing returns of naive scaling laws. Datology's mission is to automate this complex curation process, making state-of-the-art data accessible to any organization and enabling a new paradigm of AI development where data efficiency, not just raw scale, drives progress. Timestamps 00:00 Introduction 00:46 What is Datology? The mission to train models faster, better, and smaller through data curation. 01:59 Ari's background: From neuroscience to realizing the "Bitter Lesson" of AI. 05:30 Key Insight: Inductive biases from architecture become less important and even harmful as data scale increases. 08:08 Thesis: Data is the most underinvested area of AI research relative to its impact. 10:15 Why data work is culturally undervalued in research and industry. 12:19 How self-supervised learning changed everything, moving from a data-scarce to a data-abundant regime. 17:05 Why automated curation is superior to human-in-the-loop, citing the DCLM study. 19:22 The "Elephants vs. Dogs" analogy for managing data redundancy and complexity. 22:46 A brief history and commentary on key datasets (Common Crawl, GitHub, Books3). 26:24 Breaking naive scaling laws by improving data quality to maintain high marginal information gain. 29:07 Datology's demonstrated impact: Achieving baseline performance 12x faster. 34:19 The business of data: Datology's moat and its relationship with open-source datasets. 39:12 Synthetic Data Explain ed: The difference between risky "net-new" creation and powerful "rephrasing." 49:02 The Resurgence of Curriculum Learning: Why ordering data matters in the underfitting regime. 52:55 The Future of Training: Optimizing pre-training data to make post-training more effective. 54:49 Who is training their own models and why (Sovereign AI, large enterprises). 57:24 "Train Smaller": Why inference cost makes smaller, specialized models the ultimate goal for enterprises. 01:00:19 The problem with model pruning and why data-side solutions are complementary. 01:03:03 On finding the smallest possible model for a given capability. 01:06:49 Key learnings from the RC foundation model collaboration, proving that data curation "stacks." 01:09:46 Lightning Round: What data everyone wants & who should work at Datology. 01:14:24 Commentary on Meta's superintelligence efforts and Yann LeCun's role.

Jeff Huber of Chroma joins us to talk about what actually matters in vector databases in 2025, why “modern search for AI” is different, and how to ship systems that don’t rot as context grows. Full show notes: https://www.latent.space/p/chroma 00:00 Introductions 00:48 Why Build Chroma 02:55 Information Retrieval vs. Search 04:29 Staying Focused in a Competitive AI Market 08:08 Building Chroma Cloud 12:15 Context Engineering and the Problems with RAG 16:11 Context Rot 21:49 Prioritizing Context Quality 27:02 Code Indexing and Retrieval Strategies 32:04 Chunk Rewriting and Query Optimization for Code 34:07 Transformer Architecture Evolution and Retrieval Systems 38:06 Memory as a Benefit of Context Engineering 40:13 Structuring AI Memory and Offline Compaction 45:46 Lessons from Previous Startups and Building with Purpose 47:32 Religion and Values in Silicon Valley 50:18 Company Culture, Design, and Brand Consistency 52:36 Hiring at Chroma: Designers, Researchers, and Engineers

Greg Brockman, co-founder and president of OpenAI, joins us to talk about GPT-5 and GPT-OSS, the future of software engineering, why reinforcement learning is still scaling, and how OpenAI is planning to get to AGI. 00:00 Introductions 01:04 The Evolution of Reasoning at OpenAI 04:01 Online vs Offline Learning in Language Models 06:44 Sample Efficiency and Human Curation in Reinforcement Learning 08:16 Scaling Compute and Supercritical Learning 13:21 Wall clock time limitations in RL and real-world interactions 16:34 Experience with ARC Institute and DNA neural networks 19:33 Defining the GPT-5 Era 22:46 Evaluating Model Intelligence and Task Difficulty 25:06 Practical Advice for Developers Using GPT-5 31:48 Model Specs 37:21 Challenges in RL Preferences (e.g., try/catch) 39:13 Model Routing and Hybrid Architectures in GPT-5 43:58 GPT-5 pricing and compute efficiency improvements 46:04 Self-Improving Coding Agents and Tool Usage 49:11 On-Device Models and Local vs Remote Agent Systems 51:34 Engineering at OpenAI and Leveraging LLMs 54:16 Structuring Codebases and Teams for AI Optimization 55:27 The Value of Engineers in the Age of AGI 58:42 Current state of AI research and lab diversity 01:01:11 OpenAI’s Prioritization and Focus Areas 01:03:05 Advice for Founders: It's Not Too Late 01:04:20 Future outlook and closing thoughts 01:04:33 Time Capsule to 2045: Future of Compute and Abundance 01:07:07 Time Capsule to 2005: More Problems Will Emerge

Chapters 00:00:00 Welcome and Guest Introduction 00:01:18 Tulu, OVR, and the RLVR Journey 00:03:40 Industry Approaches to Post-Training and Preference Data 00:06:08 Understanding RLVR and Its Impact 00:06:18 Agents, Tool Use, and Training Environments 00:10:34 Open Data, Human Feedback, and Benchmarking 00:12:44 Chatbot Arena, Sycophancy, and Evaluation Platforms 00:15:42 RLHF vs RLVR: Books, Algorithms, and Future Directions 00:17:54 Frontier Models: Reasoning, Hybrid Models, and Data 00:22:11 Search, Retrieval, and Emerging Model Capabilities 00:29:23 Tool Use, Curriculum, and Model Training Challenges 00:38:06 Skills, Planning, and Abstraction in Agent Models 00:46:50 Parallelism, Verifiers, and Scaling Approaches 00:54:33 Overoptimization and Reward Design in RL 01:02:27 Open Models, Personalization, and the Model Spec 01:06:50 Open Model Ecosystem and Infrastructure 01:13:05 Meta, Hardware, and the Future of AI Competition 01:15:42 Building an Open DeepSeek and Closing Thoughts We first had Nathan on to give us his RLHF deep dive when he was joining AI2, and now he’s back to help us catch up on the evolution to RLVR (Reinforcement Learning with Verifiable Rewards), first proposed in his Tulu 3 paper. While RLHF remains foundational, RLVR has emerged as a powerful approach for training models on tasks with clear success criteria and using verifiable, objective functions as reward signals—particularly useful in domains like math, code correctness, and instruction-following. Instead of relying solely on subjective human feedback, RLVR leverages deterministic signals to guide optimization, making it more scalable and potentially more reliable across many domains. However, he notes that RLVR is still rapidly evolving, especially regarding how it handles tool use and multi-step reasoning. We also discussed the Tulu model series, a family of instruction-tuned open models developed at AI2. Tulu is designed to be a reproducible, state-of-the-art post-training recipe for the open community. Unlike frontier labs like OpenAI or Anthropic, which rely on vast and often proprietary datasets, Tulu aims to distill and democratize best practices for instruction and preference tuning. We are impressed with how small eval suites, careful task selection, and transparent methodology can rival even the best proprietary models on specific benchmarks. One of the most fascinating threads is the challenge of incorporating tool use into RL frameworks. Lambert highlights that while you can prompt a model to use tools like search or code execution, getting the model to reliably learn when and how to use them through RL is much harder. This is compounded by the difficulty of designing reward functions that avoid overoptimization—where models learn to “game” the reward signal rather than solve the underlying task. This is particularly problematic in code generation, where models might reward hack unit tests by inserting pass statements instead of correct logic. As models become more agentic and are expected to plan, retrieve, and act across multiple tools, reward design becomes a critical bottleneck. Other topics covered: - The evolution from RLHF (Reinforcement Learning from Human Feedback) to RLVR (Reinforcement Learning from Verifiable Rewards) - The goals and technical architecture of the Tulu models, including the motivation to open-source post-training recipes - Challenges of tool use in RL: verifiability, reward design, and scaling across domains - Evaluation frameworks and the role of platforms like Chatbot Arena and emerging “arena”-style benchmarks - The strategic tension between hybrid reasoning models and unified reasoning models at the frontier - Planning, abstraction, and calibration in reasoning agents and why these concepts matter - The future of open-source AI models, including DeepSeek, OLMo, and the potential for an “American DeepSeek” - The importance of model personality, character tuning, and the model spec paradigm - Overoptimization in RL settings and how it manifests in different domains (control tasks, code, math) - Industry trends in inference-time scaling and model parallelism Finally, the episode closes with a vision for the future of open-source AI. Nathan has now written up his ambition to build an “American DeepSeek”—a fully open, end-to-end reasoning-capable model with transparent training data, tools, and infrastructure. He emphasizes that open-source AI is not just about weights; it’s about releasing recipes, evaluations, and methods that lower the barrier for everyone to build and understand cutting-edge systems. It would seem the

ChatGPT handles 2.5B prompts/day and is on track to match Google's daily searches by end of 2026. AI agents don't browse like us—they crave queryable, chunkable data for tools like ChatGPT & Perplexity. A new industry is being born, some are calling it AI SEO, others GEO, but what is clear is that it drives amazing results. Businesses are seeing 2-4x higher conversion from visitors coming from AI compared to traditional search. Robert McCloy is the co-founder of Scrunch AI (https://scrunchai.com/), a fast growing company that helps brands and businesses re-write their content on the fly based on what agents are looking for.

Saoud Rizwan and Pash from Cline joined us to talk about why fast apply models got bitter lesson'd, how they pioneered the plan + act paradigm for coding, and why non-technical people use IDEs to do marketing and generate slides. Full writeup: https://www.latent.space/p/cline X: https://x.com/latentspacepod Chapters: 00:00 - Introductions 01:35 - Plan and Act Paradigm 05:37 - Model Evaluation and Early Development of Cline 08:14 - Use Cases of Cline Beyond Coding 09:09 - Why Cline is a VS Code Extension and Not a Fork 12:07 - Economic Value of Programming Agents 16:07 - Early Adoption for MCPs 19:35 - Local vs Remote MCP Servers 22:10 - Anthropic's Role in MCP Registry 22:49 - Most Popular MCPs and Their Use Cases 25:26 - Challenges and Future of MCP Monetization 27:32 - Security and Trust Issues with MCPs 28:56 - Alternative History Without MCP 29:43 - Market Positioning of Coding Agents and IDE Integration Matrix 32:57 - Visibility and Autonomy in Coding Agents 35:21 - Evolving Definition of Complexity in Programming Tasks 38:16 - Forks of Cline and Open Source Regrets 40:07 - Simplicity vs Complexity in Agent Design 46:33 - How Fast Apply Got Bitter Lesson'd 49:12 - Cline's Business Model and Bring-Your-Own-API-Key Approach 54:18 - Integration with OpenRouter and Enterprise Infrastructure 55:32 - Impact of Declining Model Costs 57:48 - Background Agents and Multi-Agent Systems 1:00:42 - Vision and Multi-Modalities 1:01:07 - State of Context Engineering 1:07:37 - Memory Systems in Coding Agents 1:10:14 - Standardizing Rules Files Across Agent Tools 1:11:16 - Cline's Personality and Anthropomorphization 1:12:55 - Hiring at Cline and Team Culture

Speak (https://speak.com) may not be very well known to native English speakers, but they have come from a slow start in 2016 to emerge as one of the favorite partners of OpenAI, with their Startup Fund leading and joining their Series B and C as one of the new AI-native unicorns, noting that “Speak has the potential to revolutionize not just language learning, but education broadly”. Today we speak with Speak’s CTO, Andrew Hsu, on the journey of building the “3rd generation” of language learning software (with Rosetta Stone being Gen 1, and Duolingo being Gen 2). Speak’s premise is that speech and language models can now do what was previously only possible with human tutors—provide fluent, responsive, and adaptive instruction—and this belief has shaped its product and company strategy since its early days. https://www.linkedin.com/in/adhsu/ https://speak.com One of the most interesting strategic decisions discussed in the episode is Speak’s early focus on South Korea. While counterintuitive for a San Francisco-based startup, the decision was influenced by a combination of market opportunity and founder proximity via a Korean first employee. South Korea’s intense demand for English fluency and a highly competitive education market made it a proving ground for a deeply AI-native product. By succeeding in a market saturated with human-based education solutions, Speak validated its model and built strong product-market fit before expanding to other Asian markets and eventually, globally. The arrival of Whisper and GPT-based LLMs in 2022 marked a turning point for Speak. Suddenly, capabilities that were once theoretical—real-time feedback, semantic understanding, conversational memory—became technically feasible. Speak didn’t pivot, but rather evolved into its second phase: from a supplemental practice tool to a full-featured language tutor. This transition required significant engineering work, including building custom ASR models, managing latency, and integrating real-time APIs for interactive lessons. It also unlocked the possibility of developing voice-first, immersive roleplay experiences and a roadmap to real-time conversational fluency. To scale globally and support many languages, Speak is investing heavily in AI-generated curriculum and content. Instead of manually scripting all lessons, they are building agents and pipelines that can scaffold curriculum, generate lesson content, and adapt pedagogically to the learner. This ties into one of Speak’s most ambitious goals: creating a knowledge graph that captures what a learner knows and can do in a target language, and then adapting the course path accordingly. This level-adjusting tutor model aims to personalize learning at scale and could eventually be applied beyond language learning to any educational domain. Finally, the conversation touches on the broader implications of AI-powered education and the slow real-world adoption of transformative AI technologies. Despite the capabilities of GPT-4 and others, most people’s daily lives haven’t changed dramatically. Speak sees itself as part of the generation of startups that will translate AI’s raw power into tangible consumer value. The company is also a testament to long-term conviction—founded in 2016, it weathered years of slow growth before AI caught up to its vision. Now, with over $50M ARR, a growing B2B arm, and plans to expand across languages and learning domains, Speak represents what AI-native education could look like in the next decade.

When the first video diffusion models started emerging, they were little more than just “moving pictures” - still frames extended a few seconds in either direction in time. There was a ton of excitement about OpenAI’s Sora on release through 2024, but so far only Sora-lite has been widely released. Meanwhile, other good videogen models like Genmo Mochi, Pika, MiniMax T2V, Tencent Hunyuan Video, and Kuaishou’s Kling have emerged, but the reigning king this year seems to be Google’s Veo 3, which for the first time has added native audio generation into their model capabilities, eliminating the need for a whole class of lipsynching tooling and SFX editing. The rise of Veo 3 unlocks a whole new category of AI Video creators that many of our audience may not have been exposed to, but is undeniably effective and important particularly in the “kids” and “brainrot” segments of the global consumer internet platforms like Tiktok, YouTube and Instagram. By far the best documentarians of these trends for laypeople are Olivia and Justine Moore, both partners at a16z, who not only collate the best examples from all over the web, but dabble in video creation themselves to put theory into practice. We’ve been thinking of dabbling in AI brainrot on a secondary channel for Latent Space, so we wanted to get the braindump from the Moore twins on how to make a Latent Space Brainrot channel. Jump on in!

Our last AI PhD grad student feature was Shunyu Yao, who happened to focus on Language Agents for his thesis and immediately went to work on them for OpenAI. Our pick this year is Jack Morris, who bucks the “hot” trends by -not- working on agents, benchmarks, or VS Code forks, but is rather known for his work on the information theoretic understanding of LLMs, starting from embedding models and latent space representations (always close to our heart). Jack is an unusual combination of doing underrated research but somehow still being to explain them well to a mass audience, so we felt this was a good opportunity to do a different kind of episode going through the greatest hits of a high profile AI PhD, and relate them to questions from AI Engineering. Papers and References made AI grad school: https://x.com/jxmnop/status/1933884519557353716A new type of information theory: https://x.com/jxmnop/status/1904238408899101014EmbeddingsText Embeddings Reveal (Almost) As Much As Text: https://arxiv.org/abs/2310.06816Contextual document embeddings https://arxiv.org/abs/2410.02525Harnessing the Universal Geometry of Embeddings: https://arxiv.org/abs/2505.12540Language modelsGPT-style language models memorize 3.6 bits per param: https://x.com/jxmnop/status/1929903028372459909Approximating Language Model Training Data from Weights: https://arxiv.org/abs/2506.15553https://x.com/jxmnop/status/1936044666371146076LLM Inversion"There Are No New Ideas In AI.... Only New Datasets"https://x.com/jxmnop/status/1910087098570338756https://blog.jxmo.io/p/there-are-no-new-ideas-in-ai-onlymisc reference: https://junyanz.github.io/CycleGAN/ — for others hiring AI PhDs, Jack also wanted to shout out his coauthor Zach Nussbaum, his coauthor on Nomic Embed: Training a Reproducible Long Context Text Embedder.

Solving Poker and Diplomacy, Debating RL+Reasoning with Ilya, what's *wrong* with the System 1/2 analogy, and where Test-Time Compute hits a wall Timestamps 00:00 Intro – Diplomacy, Cicero & World Championship 02:00 Reverse Centaur: How AI Improved Noam’s Human Play 05:00 Turing Test Failures in Chat: Hallucinations & Steerability 07:30 Reasoning Models & Fast vs. Slow Thinking Paradigm 11:00 System 1 vs. System 2 in Visual Tasks (GeoGuessr, Tic-Tac-Toe) 14:00 The Deep Research Existence Proof for Unverifiable Domains 17:30 Harnesses, Tool Use, and Fragility in AI Agents 21:00 The Case Against Over-Reliance on Scaffolds and Routers 24:00 Reinforcement Fine-Tuning and Long-Term Model Adaptability 28:00 Ilya’s Bet on Reasoning and the O-Series Breakthrough 34:00 Noam’s Dev Stack: Codex, Windsurf & AGI Moments 38:00 Building Better AI Developers: Memory, Reuse, and PR Reviews 41:00 Multi-Agent Intelligence and the “AI Civilization” Hypothesis 44:30 Implicit World Models and Theory of Mind Through Scaling 48:00 Why Self-Play Breaks Down Beyond Go and Chess 54:00 Designing Better Benchmarks for Fuzzy Tasks 57:30 The Real Limits of Test-Time Compute: Cost vs. Time 1:00:30 Data Efficiency Gaps Between Humans and LLMs 1:03:00 Training Pipeline: Pretraining, Midtraining, Posttraining 1:05:00 Games as Research Proving Grounds: Poker, MTG, Stratego 1:10:00 Closing Thoughts – Five-Year View and Open Research Directions

Chris Lattner of Modular (https://modular.com) joined us (again!) to talk about how they are breaking the CUDA monopoly, what it took to match NVIDIA performance with AMD, and how they are building a company of "elite nerds". X: https://x.com/latentspacepod Substack: https://latent.space 00:00:00 Introductions 00:00:12 Overview of Modular and the Shape of Compute 00:02:27 Modular’s R&D Phase 00:06:55 From CPU Optimization to GPU Support 00:11:14 MAX: Modular’s Inference Framework 00:12:52 Mojo Programming Language 00:18:25 MAX Architecture: From Mojo to Cluster-Scale Inference 00:29:16 Open Source Contributions and Community Involvement 00:32:25 Modular's Differentiation from VLLM and SGLang 00:41:37 Modular’s Business Model and Monetization Strategy 00:53:17 DeepSeek’s Impact and Low-Level GPU Programming 01:00:00 Inference Time Compute and Reasoning Models 01:02:31 Personal Reflections on Leading Modular 01:08:27 Daily Routine and Time Management as a Founder 01:13:24 Using AI Coding Tools and Staying Current with Research 01:14:47 Personal Projects and Work-Life Balance 01:17:05 Hiring, Open Source, and Community Engagement

Emmanuel Amiesen is lead author of “Circuit Tracing: Revealing Computational Graphs in Language Models” (https://transformer-circuits.pub/2025/attribution-graphs/methods.html ), which is part of a duo of MechInterp papers that Anthropic published in March (alongside https://transformer-circuits.pub/2025/attribution-graphs/biology.html ). We recorded the initial conversation a month ago, but then held off publishing until the open source tooling for the graph generation discussed in this work was released last week: https://www.anthropic.com/research/open-source-circuit-tracing This is a 2 part episode - an intro covering the open source release, then a deeper dive into the paper — with guest host Vibhu Sapra (https://x.com/vibhuuuus ) and Mochi the MechInterp Pomsky (https://x.com/mochipomsky ). Thanks to Vibhu for making this episode happen! While the original blogpost contained some fantastic guided visualizations (which we discuss at the end of this pod!), with the notebook and Neuronpedia visualization (https://www.neuronpedia.org/gemma-2-2b/graph ) released this week, you can now explore on your own with Neuronpedia, as we show you in the video version of this pod.

Solomon most famously created Docker and now runs Dagger… which has something special to share with you on Thursday. Catch Dagger at: - Tuesday: Dagger’s workshop https://www.ai.engineer/schedule#ship-agents-that-ship-a-hands-on-workshop-for-swe-agent-builders - Wednesday: Dagger’s talk: https://www.ai.engineer/schedule#how-to-trust-an-agent-with-software-delivery - Thursday: Solomon’s Keynote https://www.ai.engineer/schedule#containing-agent-chaos

As part of our AI Engineer World’s Fair preview, we’re releasing a special cross podcast recorded with Sam Charrington of TWiML AI at last week’s Google I/O! TUESDAY: Shrestha and Kwindla’s workshop: https://www.ai.engineer/schedule#milliseconds-to-magic-real-time-workflows-using-the-gemini-live-api-and-pipecat TUESDAY: Kwindla’s workshop: https://www.ai.engineer/schedule#building-voice-agents-with-gemini-and-pipecat WEDNESDAY: Shrestha and Kwindla’s talk: https://www.ai.engineer/schedule#milliseconds-to-magic-real-time-workflows-using-the-gemini-live-api-and-pipecat WEDNESDAY: Kwindla’s keynote: https://www.ai.engineer/schedule#-voice-keynote-your-realtime-ai-is-ngmi THURSDAY: Logan’s keynote: https://www.ai.engineer/schedule#a-year-of-gemini-progress-what-comes-next Catch all the speakers at AIE (both workshops and talks): Logan Kilpatrick: https://www.latent.space/p/chatgpt-gpt4-hype-and-building-llm Shrestha Basu Mallick: https://www.linkedin.com/in/shresthabm/ Kwindla Hultman Kramer: https://www.linkedin.com/in/kwkramer

One of the new tracks at next week’s AI Engineer conference in SF is a new focus on LLMs + Robotics, ft. household names like Waymo and Physical Intelligence. However there are many other companies applying LLMs and VLMs in the real world! CloudChef, the first industrial-scale kitchen robotics company with one-shot demonstration learning and an incredibly simple business model, will be serving tasty treats all day with Zippy (https://www.cloudchef.co/zippy ) their AI Chef platform. This is a lightning pod with CEO Nikhil Abraham to preview what Zippy is capable of! https://www.cloudchef.co/platform See a real chef comparison: https://www.youtube.com/watch?v=INDhZ7LwSeo&t=64s See it in the AI Engineer Expo at SF next week: https://ai.engineer

We are joined by Eno Reyes and Matan Grinberg, the co-founders of Factory.ai. They are building droids for autonomous software engineering, handling everything from code generation to incident response for production outages. After raising a $15M Series A from Sequoia, they just released their product in GA! https://factory.ai/ https://x.com/latentspacepod

In an otherwise heavy week packed with Microsoft Build, Google I/O, and OpenAI io, the worst kept secret in biglab land was the launch of Claude 4, particularly the triumphant return of Opus, which many had been clamoring for. We will leave the specific Claude 4 recap to AINews, however we think that both Gemini’s progress on Deep Think this week and Claude 4 represent the next frontier of progress on inference time compute/reasoning (at last until GPT5 ships this summer). Will Brown’s talk at AIE NYC and open source work on verifiers have made him one of the most prominent voices able to publicly discuss (aka without the vaguepoasting LoRA they put on you when you join a biglab) the current state of the art in reasoning models and where current SOTA research directions lead. We discussed his latest paper on Reinforcing Multi-Turn Reasoning in LLM Agents via Turn-Level Credit Assignment and he has previewed his AIEWF talk on Agentic RL for those with the temerity to power thru bad meetup audio.

ChatGPT Codex is here - the first cloud hosted Autonomous Software Engineer (A-SWE) from OpenAI. We sat down for a quick pod with two core devs on the ChatGPT Codex team: Josh Ma and Alexander Embiricos to get the inside scoop on the origin story of Codex, from WHAM to its future roadmap. Follow them: https://github.com/joshma and https://x.com/embirico Chapters - 00:00 Introduction to the Latent Space Podcast - 00:59 The Launch of ChatGPT Codex - 03:08 Personal Journeys into AI Development - 05:50 The Evolution of Codex and AI Agents - 08:55 Understanding the Form Factor of Codex - 11:48 Building a Software Engineering Agent - 14:53 Best Practices for Using AI Agents - 17:55 The Importance of Code Structure for AI - 21:10 Navigating Human and AI Collaboration - 23:58 Future of AI in Software Development - 28:18 Planning and Decision-Making in AI Development - 31:37 User, Developer, and Model Dynamics - 35:28 Building for the Future: Long-Term Vision - 39:31 Best Practices for Using AI Tools - 42:32 Understanding the Compute Platform - 48:01 Iterative Deployment and Future Improvements

More info: https://docs.anthropic.com/en/docs/claude-code/overview The AI coding wars have now split across four battlegrounds: 1. AI IDEs: with two leading startups in Windsurf ($3B acq. by OpenAI) and Cursor ($9B valuation) and a sea of competition behind them (like Cline, Github Copilot, etc). 2. Vibe coding platforms: Bolt.new, Lovable, v0, etc. all experiencing fast growth and getting to the tens of millions of revenue in months. 3. The teammate agents: Devin, Cosine, etc. Simply give them a task, and they will get back to you with a full PR (with mixed results) 4. The cli-based agents: after Aider’s initial success, we are now seeing many other alternatives including two from the main labs: OpenAI Codex and Claude Code. The main draw is that 1) they are composable 2) they are pay as you go based on tokens used. Since we covered all three of the first categories, today’s guests are Boris and Cat, the lead engineer and PM for Claude Code. If you only take one thing away from this episode, it’s this piece from Boris: Claude Code is not a product as much as it’s a Unix utility. This fits very well with Anthropic’s product principle: “do the simple thing first.” Whether it’s the memory implementation (a markdown file that gets auto-loaded) or the approach to prompt summarization (just ask Claude to summarize), they always pick the smallest building blocks that are useful, understandable, and extensible. Even major features like planning (“/think”) and memory (#tags in markdown) fit the same idea of having text I/O as the core interface. This is very similar to the original UNIX design philosophy: Claude Code is also the most direct way to consume Sonnet for coding, rather than going through all the hidden prompting and optimization than the other products do. You will feel that right away, as the average spend per user is $6/day on Claude Code compared to $20/mo for Cursor, for example. Apparently, there are some engineers inside of Anthropic that have spent >$1,000 in one day! If you’re building AI developer tools, there’s also a lot of alpha on how to design a cli tool, interactive vs non-interactive modes, and how to balance feature creation. Enjoy! Timestamps [00:00:00] Intro [00:01:59] Origins of Claude Code [00:04:32] Anthropic’s Product Philosophy [00:07:38] What should go into Claude Code? [00:09:26] Claude.md and Memory Simplification [00:10:07] Claude Code vs Aider [00:11:23] Parallel Workflows and Unix Utility Philosophy [00:12:51] Cost considerations and pricing model [00:14:51] Key Features Shipped Since Launch [00:16:28] Claude Code writes 80% of Claude Code [00:18:01] Custom Slash Commands and MCP Integration [00:21:08] Terminal UX and Technical Stack [00:27:11] Code Review and Semantic Linting [00:28:33] Non-Interactive Mode and Automation [00:36:09] Engineering Productivity Metrics [00:37:47] Balancing Feature Creation and Maintenance [00:41:59] Memory and the Future of Context [00:50:10] Sandboxing, Branching, and Agent Planning [01:01:43] Future roadmap [01:11:00] Why Anthropic Excels at Developer Tools

Note from your hosts: we were off this week for ICLR and RSA! This week we’re bringing you one of the top episodes from our lightning podcast series, the shorter format, Youtube-only side podcast we do for breaking news and faster turnaround. Please support our work on YouTube! https://www.youtube.com/playlist?list=PLWEAb1SXhjlc5qgVK4NgehdCzMYCwZtiB The explosion of embedding-based applications created a new challenge: efficiently storing, indexing, and searching these high-dimensional vectors at scale. This gap gave rise to the vector database category, with companies like Pinecone leading the charge in 2022-2023 by defining specialized infrastructure for vector operations. The category saw explosive growth following ChatGPT's launch in late 2022, as developers rushed to build AI applications using Retrieval-Augmented Generation (RAG). This surge was partly driven by a widespread misconception that embedding-based similarity search was the only viable method for retrieving context for LLMs!!! The resulting "vector database gold rush" saw massive investment and attention directed toward vector search infrastructure, even though traditional information retrieval techniques remained equally valuable for many RAG applications. https://x.com/jobergum/status/1872923872007217309 Chapters 00:00 Introduction to Trondheim and Background 03:03 The Rise and Fall of Vector Databases 06:08 Convergence of Search Technologies 09:04 Embeddings and Their Importance 12:03 Building Effective Search Systems 15:00 RAG Applications and Recommendations 17:55 The Role of Knowledge Graphs 20:49 Future of Embedding Models and Innovations

Vasek Mlejnsky from E2B joins us today to talk about sandboxes for AI agents. In the last 2 years, E2B has grown from a handful of developers building on it to being used by ~50% of the Fortune 500 and generating millions of sandboxes each week for their customers. As the “death of chat completions” approaches, LLMs workflows and agents are relying more and more on tool usage and multi-modality. The most common use cases for their sandboxes: - Run data analysis and charting (like Perplexity) - Execute arbitrary code generated by the model (like Manus does) - Running evals on code generation (see LMArena Web) - Doing reinforcement learning for code capabilities (like HuggingFace) Timestamps: 00:00:00 Introductions 00:00:37 Origin of DevBook -> E2B 00:02:35 Early Experiments with GPT-3.5 and Building AI Agents 00:05:19 Building an Agent Cloud 00:07:27 Challenges of Building with Early LLMs 00:10:35 E2B Use Cases 00:13:52 E2B Growth vs Models Capabilities 00:15:03 The LLM Operating System (LLMOS) Landscape 00:20:12 Breakdown of JavaScript vs Python Usage on E2B 00:21:50 AI VMs vs Traditional Cloud 00:26:28 Technical Specifications of E2B Sandboxes 00:29:43 Usage-based billing infrastructure 00:34:08 Pricing AI on Value Delivered vs Token Usage 00:36:24 Forking, Checkpoints, and Parallel Execution in Sandboxes 00:39:18 Future Plans for Toolkit and Higher-Level Agent Frameworks 00:42:35 Limitations of Chat-Based Interfaces and the Future of Agents 00:44:00 MCPs and Remote Agent Capabilities 00:49:22 LLMs.txt, scrapers, and bad AI bots 00:53:00 Manus and Computer Use on E2B 00:55:03 E2B for RL with Hugging Face 00:56:58 E2B for Agent Evaluation on LMArena 00:58:12 Long-Term Vision: E2B as Full Lifecycle Infrastructure for LLMs 01:00:45 Future Plans for Hosting and Deployment of LLM-Generated Apps 01:01:15 Why E2B Moved to San Francisco 01:05:49 Open Roles and Hiring Plans at E2B

We’ll keep this brief because we’re on a tight turnaround: GPT 4.1, previously known as the Quasar and Optimus models, is now live as the natural update for 4o/4o-mini (and the research preview of GPT 4.5). Though it is a general purpose model family, the headline features are: Coding abilities (o1-level SWEBench and SWELancer, but ok Aider) Instruction Following (with a very notable prompting guide) Long Context up to 1m tokens (with new MRCR and Graphwalk benchmarks) Vision (simply o1 level) Cheaper Pricing (cheaper than 4o, greatly improved prompt caching savings) We caught up with returning guest Michelle Pokrass and Josh McGrath to get more detail on each!