GitHub

Exo: Harnesses should see their own code and logs — Alex Krentsel, UC Berekeley / Google Research

Aug 15, 2026 · 47:11

Alex Krentsel presents Exo, a fully recursive self-improving agent harness he built with Martín Casado and Anker Goyal, arguing agents can now safely edit their own code and runtime policy. Exo's architecture splits an agent into a stateless executor, a stateful harness with history and secrets, and an isolated sandbox, letting it change every component, roll back automatically, and teleport state—unlike OpenClaw/Pi, which only allow memory, skills, and tool extensions. The episode details Exo cutting its Discord adapter costs by 96% at runtime, plus secret stores and evals to prevent reward hacking. Krentsel argues RSI is newly possible because the harness is code in the same medium as LLM output; Exo runs in production at Brain Trust, with contributors directed to github.com/exoharness/exo.

The Blueprint for Autonomous Work Agents | Gavriel Cohen, NanoClaw

Jun 29, 2026 · 23:15

Gavriel Cohen, founder of NanoClaw, explains why he built a minimal, secure alternative to OpenClaw after feeling apprehensive about its codebase and dependencies for production use. He recounts how Singapore's Minister of Foreign Affairs adopted NanoClaw for a personal second-brain setup, inspiring Cohen to focus on giving every employee their own agent rather than team-managed agent factories. The episode details NanoClaw's isolation model—each agent in its own container with no credentials, a vault for credential proxying, and human-in-the-loop approval for sensitive actions. Cohen also shares the pivot from an AI-native marketing agency to nanoCo, which now deploys and manages agents for over 100 enterprise clients, and discusses the challenge of triaging AI-generated pull requests in open source.

Satya Nadella on AI: @NoPriorsPodcast x Latent Space Crossover Special at Microsoft Build 2026

Jun 3, 2026 · 41:27

Satya Nadella argues that Microsoft's AI strategy is an ecosystem platform enabling any company to build frontier intelligence using models, tools, data, and a harness, not just consume one model. He outlines MAI training with clean data lineage, hill-climbing scaffolds, and private evals as core IP. The harness concept features multi-model harnesses with strong context layers, exemplified by GitHub Copilot and Work IQ turning M365 data into a database for agents. He notes coding agents required new IDE/UI, and long-running autopilots create value. Pricing evolves from per-user to consumption, and SaaS will unbundle and rebundle. Engineering generalists gain leverage; infrastructure roles like RLEs remain critical. Datacenter expansion requires community permission, with benefits in healthcare and rethinking education. Microsoft built more Azure capacity in 15 months than its first 15 years.

Notion’s Sarah Sachs & Simon Last on Custom Agents, Evals, and the Future of Work

Apr 15, 2026 · 1:25:37

Sarah Sachs and Simon Last of Notion explain how Custom Agents were rebuilt four to five times since 2022, with early attempts failing due to lack of tool-calling standards and short context windows. They shifted from few-shot prompting to tool definitions and progressive disclosure, now supporting over 100 tools. Their eval system includes frontier evals that pass only 30% to gauge model progress, and they employ Model Behavior Engineers to write evals and triage failures. The episode covers their views on MCP versus CLI, meeting notes as data capture, and the software factory concept of agents collaborating to develop codebases. They also discuss Notion's culture of low-ego, high-velocity rebuilding and the philosophy that every surface must work for both humans and agents.

Extreme Harness Engineering: 1M LOC, 1B toks/day, 0% human code or review — Ryan Lopopolo, OpenAI

Apr 7, 2026 · 1:17:54

Ryan Lopopolo of OpenAI's Frontier team details his team's extreme experiment in harness engineering: building a 1M LOC internal Electron app with zero human-written or reviewed code, relying instead on Codex agents that process 1B tokens daily ($2-3k/day). He argues that humans are the bottleneck and that teams should encode non-functional requirements into specs, skills, and observability tooling rather than prompting agents to 'try harder.' The result is Symphony, a ghost library spec and Elixir reference implementation that automates the entire pull-request lifecycle, including self-review and merge. Lopopolo explains how the team scales by treating software as agent-legible text, using worktrees for multi-agent collaboration, and feeding agent mistakes back into the repository via docs like core-beliefs.md. He also discusses Frontier's enterprise platform for safe agent deployment, noting that success depends on giving agents full context—even company culture and inside jokes—so they can act as full teammates.

Cursor's Third Era: Cloud Agents — ft. Sam Whitmore, Jonas Nelle, Cursor

Mar 6, 2026 · 1:11:18

Cursor's Sam Whitmore and Jonas Nelle detail their Cloud Agents launch, where agents get full cloud VMs to onboard repos, run code, and test changes end-to-end before returning PRs with demo videos and remote desktop access. The model decides whether to test (with /no-test to skip and /repro for bug reproduction videos). Videos cut review time by letting humans watch 20-second clips instead of giant diffs. Cursor supports parallel model swarms via best-of-N—running multiple models head-to-head on the same prompt—and subagents for context management, plus a long-running 'grind mode' for multi-day tasks. They predict cloud agents will surpass local agents in volume by year-end, driving per-developer spend to thousands per month and creating new bottlenecks in CI/CD and production pipelines.

One Year of MCP — with David Soria Parria and AAIF leads from OpenAI, Goose, Linux Foundation

Dec 28, 2025 · 1:39:19

David Soria Parra, MCP lead at Anthropic, along with Jim Zemlin (Linux Foundation CEO), Nick Cooper (OpenAI), and Brad Howes (Block/Goose), recount the one-year evolution of the Model Context Protocol from a local experiment to the de facto standard for agentic systems, now donated to the new Agentic AI Foundation. They detail four spec releases—from local stdio to remote HTTP streaming, OAuth 2.1 authentication (and enterprise lessons learned), long-running tasks, and MCP Apps (iframes for richer UI)—and explain why internal enterprise adoption is exploding faster than expected, mostly invisible and at massive scale. The group reveals how three competitive AI labs came together to donate protocols and agents to a neutral foundation, how the foundation will balance taste-making (curating meaningful projects) with openness, and their 2025 vision: MCP as the communication layer for asynchronous, long-running agents that discover and install their own tools, unlocking another order of magnitude in AI productivity.

⚡️GPT5-Codex-Max: Training Agents with Personality, Tools & Trust — Brian Fioca + Bill Chen, OpenAI

Dec 26, 2025 · 27:46

OpenAI’s Brian Fioca and Bill Chen discuss the launch of Codex Max, a long-running coding agent designed to work 24+ hours, manage its own context, and spawn sub-agents. They reveal how GPT-5 training prioritized personality traits—communication, planning, and self-checking—to build developer trust, and how Codex develops tool habits like preferring 'rg' over 'grep'. The abstraction layer is shifting from models to full-stack agents, enabling partners to plug Codex into tools like VS Code or Zed. They highlight applied evals that measure real-world impact over academic benchmarks, and the need for multi-turn eval APIs. Their 2026 vision: coding agents trusted enough to handle the hardest refactors at any company, not just top-tier firms, and general enough to automate personal workflows like organizing desktops or email.

Amp: The Emperor Has No Clothes

Sep 25, 2025 · 1:21:59

Quinn Slack (CEO) and Thorsten Ball (Amp Dictator) from Sourcegraph discuss Amp Code, a coding agent that ships 15x/day with no code reviews. They explain the strategic pivot from Cody to Amp, emphasizing that building the best coding agent trumps everything else. The team dogfoods Amp internally, using it to build itself, and maintains both a VS Code extension and CLI, with a 50/50 internal split. They argue that model choice is becoming less important than the harness around the model, and they reject features like prompt enhancers, custom subagents, and MCP servers as not adding value. Common failure modes include outsourcing thinking but not typing and using agents without proper context engineering. They predict async agents running 24/7 will dominate, and they target early adopters willing to adapt their codebases and workflows for maximal agent effectiveness.

Better Data is All You Need — Ari Morcos, Datology

Aug 29, 2025 · 1:18:43

Ari Morcos, CEO of Datology, argues that data curation is the most impactful and underinvested area in AI, claiming models are defined by what they eat. He details his shift from focusing on inductive biases to realizing data quality is paramount, citing the DCLM study where human experts could not predict automated filtering decisions. Datology's methods combine filtering, rebalancing, curriculum, and rephrasing synthetic data to achieve baseline performance 12x faster, improve accuracy by 4–5 absolute points, and train models with fewer than half the parameters. Morcos explains how curating data bends naive scaling laws by maintaining marginal information gain, and shares results from the Arc 4.5B model, which was trained on 7 trillion tokens (down from 25 trillion) using Datology's curation. He highlights that data is a compute multiplier, making it possible for enterprises to train smaller, cheaper models for specific domains.

The AI Coding Factory

May 29, 2025 · 59:23

Factory.ai co-founders Eno Reyes and Matan Grinberg join Latent Space to explain their vision for autonomous software engineering droids that handle everything from code generation to incident response for enterprises. They met at a Langchain Hackathon in 2023, and within eight days both quit their jobs to start Factory. The platform is browser-based, not IDE-based, because they argue the optimal UI for a world where humans write less code will not evolve from the IDE. They focus on enterprise codebases, citing a large migration that took from four months to three and a half days. Key design decisions include a delegation model where agents ask clarifying questions, proactive context gathering via synthetic insights, and usage-based pricing tied to token efficiency. They also discuss why SWE-Bench is dead for their use case, the challenge of model post-training biases toward CLI tools, and their need for 'junior Enos' in go-to-market roles.

The Creators of Model Context Protocol

Apr 3, 2025 · 1:20:21

Justin Spahr-Summers and David Soria Parra, co-creators of Anthropic's Model Context Protocol (MCP), explain its origin as a solution to the M×N integration problem they experienced when copying data between Claude Desktop and IDEs. They detail MCP's design inspirations from LSP, its three core primitives (tools for model-driven actions, resources for app-controlled data, and prompts for user-initiated macros), and why they chose JSON-RPC with stateful connections. The episode contrasts MCP with OpenAPI, discusses client control and trust in servers, and outlines the future roadmap including a streamable HTTP transport and remote-server authorization via OAuth 2.1. The guests express a wish for more sampling clients and full spec support, and emphasize that MCP is an open, multi-company project with community commit access.

The #1 SWE-Bench Verified Agent

Apr 2, 2025 · 31:46

Guy Gur-Ari, co-founder of Augment Code, explains how their coding agent achieved the #1 spot on SWE-Bench Verified without fine-tuning, using off-the-shelf models with techniques like sequential thinking, ensembling, and code base understanding. He shares that orientation and memories help the agent learn codebase conventions, and that MCP support allows extensibility. Gur-Ari contrasts Augment's IDE-first approach with competitors like Devin and Factory, emphasizing meeting developers where they are. He also discusses the future of IDEs, the role of custom models, and recommends the SWE-RL paper for reinforcement learning in coding. The episode features a live demo of the agent implementing a feature from a Linear ticket and creating a PR.

OpenAI o1 isn’t a chat model (and that’s the point)

Jan 17, 2025 · 31:59

Ben Hylak and Dan McAteer explain why OpenAI’s o1 is not a chat model and how mastering its prompting turned Ben from a skeptic into a proponent. Ben’s prompt structure—goal, return format, warnings, context dump—emphasizes describing what you want rather than how to think, a shift from earlier models. Dan uses o1 for coding by feeding it full code context, achieving one-shot implementations that reach 100% where previous models maxed out at 95%. Ben notes o1 is the most capable but hardest model to use, with experiments costing $20 per try, and argues model routing will let humans or systems trade off cost, speed, and intelligence. Dan also applies o1 to scientific research, feeding it papers from multiple fields to find novel connections that humans would miss.

Best of 2024 in Agents (from #1 on SWE-Bench Full, Prof. Graham Neubig of OpenHands/AllHands)

Dec 25, 2024 · 51:25

Professor Graham Neubig of CMU and AllHands AI, maintainer of the open-source coding agent framework OpenHands, delivers a talk on the state of agents in 2024, arguing that agents are now capable of automating significant portions of software development but still struggle with information gathering and error recovery. He demonstrates OpenHands solving real coding tasks live, explains key design choices like using Python code execution over granular tool calls to reduce LLM invocations, and shares that Claude currently outperforms other models for agentic tasks due to superior instruction following and error correction. Neubig discusses eight perennial agent problems including agent-computer interface, human-agent interface, model selection, planning, workflows, exploration, search, and evaluation. He predicts every major LLM trainer will focus on agent-oriented models by mid-2025, that agent benchmarks like SWE-Bench and WebArena will saturate and require harder successors, and that the biggest challenge will be designing human-agent interfaces for systems that succeed only 30-40% of the time autonomously but 80-90% with human feedback. The talk also covers workflow memory as a…

Agents @ Work: Lindy.ai (with live demo!)

Nov 15, 2024 · 1:08:01

Florent Crivello, founder of Lindy.ai, explains how his no-code agent platform evolved from an overly LLM-dependent design to a deterministic 'on rails' system that puts AI in a box, making agents more reliable and easier to use. The product was rebuilt from scratch as Lindy 2.0, featuring step-by-step workflows with structured outputs and memory management that still requires pruning. Crivello reveals that Claude 3.5 Sonnet ended model bottlenecks but warns against over-engineering cognitive architectures, advocating instead for letting models scale. He discusses building their own eval tools (now considering Braintrust), hiring vertical GMs for go-to-market, and the decision to return to in-person work, arguing remote work hampers creativity. He also shares his contrarian views on Europe vs. US tech, testing the Overton window with spicy tweets, and balancing AI safety concerns (10% p(doom)) with building toward a potential utopia.

Agents @ Work: Dust.tt — with Stanislas Polu

Nov 11, 2024 · 58:53

Stanislas Polu, co-founder of Dust.tt, argues that horizontal agent platforms—not vertical, single-use-case agents—will win the enterprise by letting non-technical employees build their own automation. A former OpenAI researcher who worked directly with Ilya Sutskever on formal mathematics and saw GPT-4 internally before ChatGPT, Polu left in 2022 to start Dust, now an open-source infrastructure layer that connects company data (Slack, Notion, GitHub) to LLMs and lets anyone create simple, instruction-driven agents. He contrasts Dust’s API-first, integration-heavy approach with LangChain’s open-source orchestration and Adept’s browser automation bet, insisting that most useful agent workflows require only a few tool calls and that function-calling quality (where GPT-4 Turbo still leads) matters more than model size. Dust achieves 88% daily active user penetration at some enterprise customers, and Polu predicts the first billion-dollar company run by a single person will emerge once platforms like Dust mature enough to let tiny teams scale to massive output.

Is finetuning GPT4o worth it?

Aug 22, 2024 · 1:01:16

Ali Pullen, founder of Cosine, discusses Genie, an AI software engineering colleague that achieved 30% on SWE-Bench and 43.8% on SWE-Bench Verified via fine-tuning GPT-4o on billions of tokens of synthetic data. Cosine built Genie by training on the process of software engineering—not just working code—using synthetic runtime errors and a self-improvement loop. The agent follows a four-stage workflow: code retrieval (66% accuracy), planning, writing diffs, and running CI tests. Pullen reveals that model performance degrades linearly beyond 60K tokens in the context window, and that OpenAI dynamically sizes LoRA adapters to handle Cosine's massive dataset. He explains why Genie isn't on the SWE-Bench leaderboard (refusing to publish trajectories to prevent distillation) and outlines plans to scale the dataset and fine-tune on customer codebases for personalized performance.

Making Transformers Sing - with Mikey Shulman of Suno

Mar 14, 2024 · 58:58

In this episode of Latent Space, hosts Alessio and Shawn Wang interview Mikey Shulman, CEO of Suno, about making transformers sing. Suno uses transformers to predict audio tokens end-to-end, avoiding baked-in musical knowledge, with a tokenization secret sauce that also includes non-music audio for better vocal realism. Their models are relatively small (far below 175B parameters) due to latency needs, and they prioritize scaling research over brute-force size. Over half of Suno users employ expert mode, tweaking lyrics and style prompts, rather than easy mode. Shulman argues Suno is not the 'Midjourney of music' because music is inherently social and synchronous, unlike images. He demoed live generation, showing control via tokens like [beat drop] and style modifiers, and revealed future plans for collaborative concerts, continuous DJ modes, and personalized models. He also advocates for hiring economists to avoid Goodhart's law pitfalls in ML benchmarks, especially in audio where aesthetics matter most.