Daily Brief

After 10 years of monthly releases, VS Code now ships weekly. It wasn't more AI that made it work. It was rebuilding the whole system around it. Harald Kirschner, principal product manager on VS Code at Microsoft, explains how a small team shipping to more than 50 million users moved from monthly to weekly releases with AI agents, and what broke along the way. He covers three stages: shipping faster (agent-ready codebases, skills, TypeScript Go's 10x faster builds, letting agents drive the app), holding quality (mandatory AI code review, AI issue triage, error stacks that turn into auto-fix P

No Mac Mini, no laptop agent. Jeremy Adams runs his personal AI agent on an old Raspberry Pi he wears around his neck. Jeremy Adams, senior developer advocate at Neo4j, assembles a wearable AI agent live on stage: NanoClaw on a Raspberry Pi 4B, messaged over WhatsApp, with Docker-isolated agents and a Neo4j graph database. He shows how he built graph memory using the POLE schema (person, object, location, event) mid-flight, added voice-to-text through a USB mic and a hardware button, ran an offline mode with a local Neo4j, and walked the AI Engineer expo floor turning booth notes into a graph

Models became agents, and the APIs didn't keep up. Google DeepMind rebuilt them. Ivan Leo, developer experience engineer at Google DeepMind, introduces the Gemini Interactions API and Managed Agents. He traces how we went from single completions to function calling to agents, then shows server-side state with interaction IDs (no more hand-managing thought signatures), strongly typed multimodal outputs, mixing built-in and custom tools, and the new steps data model. He then demos Managed Agents, which run the Antigravity harness in a persistent remote sandbox with environment IDs, loadable sou

Full episode: https://youtu.be/LwQQ7nBCGSs Me on twitter: https://x.com/dwarkesh_sp

Across thousands of GPUs, hardware failures are guaranteed. Fixing them by hand at 3 a.m. doesn't scale. Connor Guerrero, Young Jeong and Nikhil Gupta from Crusoe explain why they built Managed Slurm on top of Kubernetes. Slurm gives researchers gang scheduling, topology awareness and familiar sbatch workflows, but falls short on dynamic resources, node health and observability. Kubernetes fills those gaps without either team changing how it works. They walk through AutoClusters' fully automatic remediation of an XID 79 GPU failure, then demo killing a GPU mid-training and resuming from check

Humans don't respond in 200 milliseconds. Your agent shouldn't fall over waiting for one. Melanie Warrick from developer relations at Temporal shows how to build human-in-the-loop agents that don't block or lose state. Using a live multi-agent ice cream delivery demo with fleet, customer and dispatch agents on Google ADK and LangGraph, she explains why durability belongs in the agent harness and how Temporal's workers, workflows and activities track state. She shows how two primitives, a wait condition and a signal, let an agent pause for a human for minutes or weeks. Then she kills the worke

Shoppers are moving from typing keywords to describing what they need, and agents are doing the searching. Nixon Dinh, director of product for agentic commerce at PayPal, explains how shopping is moving from the search era to the intent era and then the delegation era, and why product catalogs built for keyword SEO aren't ready for AI agents. He breaks down keyword vs semantic search and shares the results of PayPal's experiments enriching merchant product data. Enrichment consistently improved agent recommendations and thin catalogs gained the most, but over-enriching with unstructured boile

Claude Code can make the change. But what if you have 90,000 repositories to make it in? Dan Adler, CEO of Sourcegraph, makes the case for the people who own the huge, decades-old codebases that run banks, cars and airlines. Coding agents are producing a tidal wave of code, and those codebases are starting to decay through duplicated code, drifting standards, brittle dependencies and new vulnerabilities. Agents understand code by searching, but you can't grep what you can't see. Dan argues that large-scale code visibility is infrastructure, and introduces Sourcegraph's Agentic Batch Changes f

Full episode: https://youtu.be/LwQQ7nBCGSs Me on twitter: https://x.com/dwarkesh_sp

This agent starts with zero tools. Ask it for something it can't do, and it writes the tool and uses it without restarting. Sandhya Subramani, senior developer advocate for GenAI at AWS, demos meta-tooling with the open-source Strands Agents SDK. That means agents that write, load and use their own tools at runtime, and even spin up their own sub-agents. All it takes is a system prompt describing what a good tool looks like, plus three tools: editor, shell and load_tool. She walks through agentic patterns (swarm, graph, handoff, workflow), demos a travel planner that builds its own sub-agents