Nov 10, 2025 · 35m · latent-space
⚡ Inside GitHub’s AI Revolution: Jared Palmer Reveals Agent HQ & The Future of Coding Agents
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
In this in-depth interview on Latent Space, GitHub Senior Vice President Jared Palmer discusses the launch of GitHub Agent HQ, sharing key technical insights into the evolution of coding agents, runtime sandboxing, reliability engineering, and the modernization of core developer workflows.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. How this is scored →
The hosts as informed peer 4.9 Guest teaching 3.0 Guest disagreement 1.0 The hosts pushing back 0.9
speaking balance: gold is the hosts, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 24:35 Jared calls out industry complacency on reliability
Jared bluntly asserts that AI builders are living in 'la-la land' if they don't rigorously track dropped requests, provider uptime, and error-free multi-turn agent sessions.
Hardest push from the hosts ▶ 13:37 Host challenges model picker abstraction layerThe host pushes back against traditional model-switching interfaces, arguing that loosely bound generic abstractions produce lowest-common-denominator agent behavior.
Biggest teaching moment ▶ 31:20 Jared explains Facebook stacked diffs vs Git pull requestsJared provides a masterclass on monorepo engineering, contrasting Facebook's internal Mercurial stacked commits with traditional Git PR paradigms.
The host holds their own ▶ 15:15 Host explains skills as the universal file system interfaceWhen Jared admits unfamiliarity with Anthropic skills, the host steps up to articulate how skills package tool execution into a clean, markdown-driven file system abstraction.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | The hosts as informed peer | Guest teaching | Guest disagreement | The hosts pushing back | Why |
|---|---|---|---|---|---|---|
| Scaling from Niche Tools to GitHub's Broad Ecosystem | 4 | 3 | 1 | 1 | The conversation is very friendly and collaborative. The host prompts Jared on the origin story of v0 and AI SDK while contributing historical milestones (like GPT-4 context limits and Code Interpreter). | |
| Synthetic Composite Models vs. Multi-Agent Ecosystems | 6 | 2 | 1 | 2 | The host draws on his Cognition engineering background to discuss composite models vs. model pickers, pushing the argument that model switching at the base API layer is the wrong abstraction for coding agents. | |
| Redefining Coding Agents, Tool Execution, and Skills | 6 | 2 | 1 | 1 | When Jared asks about skills versus MCPs, the host takes the floor to explain DXT and how skills serve as an LLM-friendly file-system-first interface. | |
| GitHub Agent HQ and Unified AI Developer Workflows | 3 | 3 | 0 | 0 | Jared explains the vision behind GitHub Agent HQ and the closer integration between VS Code, Azure, and GitHub. The host reacts enthusiastically with minimal challenge. | |
| Sandboxes, Dev Containers, and Repo Environment Setup | 6 | 2 | 1 | 1 | The host brings up container sandboxing setups, noting Cognition's Kubernetes pods and his previous attempt at Netlify to open-source standardized framework auto-detection. | |
| Protocol Standardization, MCPs, and Computer Use | 6 | 2 | 1 | 1 | The host educates Jared on ACP payments and shares technical details on computer use advances driven by open OCR models like DeepSeek OCR. | |
| Reliability Engineering: The Climb from 90% to 99% Accuracy | 3 | 5 | 2 | 1 | Jared explains the harsh reality of reliability engineering in AI products, asserting that many teams live in 'la-la land' regarding infra failure rates and error-free multi-turn sessions. | |
| Applying Coding Agents to Knowledge Work and Browsers | 6 | 2 | 1 | 0 | The discussion covers applying coding agents to knowledge work and browsers. The host demonstrates technical depth by describing his custom open-source tool Chrome Dump and attempts to build within Tauri. | |
| Overhauling the GitHub Homepage Experience | 4 | 6 | 1 | 1 | Jared delivers an in-depth breakdown of stacked diffs, comparing Facebook's Buck/Mercurial workflow against Git pull requests and explaining why large repos demand stacked workflows. |
Statements from this episode (14)
GitHub launches Agent HQ at the GitHub Universe conference
“And today we launched Asian HQ among other things here at universe.”
Vercel's v0 took nine months to reach $1M ARR
“That launches, and then probably like nine months later it took us like nine months to get to like a million ARR. This little team, but then the models progressed.”
Vercel's v0 added $1M MRR every 14 days after chat rewrite
“When we launched V-Zero, the chat version, or the new V-Zero, whatever you want to call it it's like, 14 days, another million MRR, 14 days, another million MRR, it was like a rocket ship after that.”
Constraining v0 to Next.js and shadcn enabled faster execution
“What was really liberating for us was actually the focus on just one stack, or one framework, whatever else was trying to do, General purpose coding agent. We were like, no, we're just going to focus on Next.js front-end and Chad CNN. And that was really, that…”
Vercel shared its post-training harness with frontier AI model labs
“We also started working with all the frontier model labs to help because it was in our Vercel's best interest. To have them be created Next.js. And also because of the post train models, and you can read about this on the Vercel blog, like we can the post trai…”
GitHub Agent HQ supports third-party agents like Claude Code and Cognition
“Switching gears to GitHub. Like we are all about model choice now and like making sure that and what's cool is that like, it's, we also have copilot, which is our harness. And co-pilot CLI, but we also have third party partners like Cloud Code and Codex and Co…”
Microsoft CoreAI unites Visual Studio, VS Code, GitHub, and Azure
“And so you think about the new core core AI organization, you've got Visual Studio Code, GitHub, and parts of Azure, all in one.”
Microsoft has competing internal runtime sandbox solutions for AI agents
“There's work and discussion about what that runtime should be even internally at Microsoft. We've got a couple of different competing things, so we'll figure it out in the next, you know, in the next cycle here.”
MCP is the primary method enterprise customers use to add agent context
“The MCP is huge, it seems like. It is the way that a lot of the, especially when it comes to, like, digital transformation or some of our enterprise customers, it's where they're going to be able to be at, or where they are able to add context.”
GitHub announces custom agent support with prompts and MCPs in Agent HQ
“In addition to that, we also have custom agents that we announced today, too. So, like, you can work with prompts and stuff within your agent HQ and customize these agents for different tasks, and those can have MCPs and such.”
AI builders are blind to poor quality without tracking error-free sessions
“Most people are blind, like living in la la land about how poor quality their AI product lately is, unless they're really measuring like the number of error free sessions, like how many errors are coming from the infra providers, like, you know, how, how many …”
Autonomous AI agents require vastly more reliable infrastructure to succeed
“Agents will really only work where we get to not only like the more intelligent models, but better reliability of the infrastructure providers, right?”
GitHub rejected stacked diff prototypes in 2020 and 2022 as too risky
“And there's been multiple attempts at this internally going back to like, 20, 20. And there was one very, very, very polished attempt to in 20, 22. And it just, I don't have all the context, so, but it was, there was a pretty good implementation. All of the wo…”
GitHub is actively exploring stacked diff support for its product roadmap
“So anyway, we're, we had cold meetings internally already, and we're trying to weave it into the planning and the roadmap, and so hopefully we'll be able to share more updates soon, but like, it's a Top of the list known feature. And so, and like, we're workin…”