Written by Stephen Simon on

First glance at Entire: rethinking where code lives in the agent era

Entire wants to rethink where agent-written code lives. Here’s our first glance at the new Git forge from former GitHub CEO Thomas Dohmke.

6 min read

Global AI Community blog banner: “First glance at Entire.” Many AI agents feeding into one human review.
Entire Video

This week brought a notable launch in developer tools. Entire, the startup led by former GitHub CEO Thomas Dohmke, released a full platform for hosting, reviewing and shipping code written by AI agents. Agents are a daily topic in our community, at AgentCon, at the Global AI Bootcamp and in our local chapters. That makes one question hard to ignore: where does all that agent-written code actually live?

This is a first glance, not a full review. We looked at what Entire announced, what it claims and what it could mean for the builders in our community. Hands-on impressions will follow once more of us have tried it.

What is Entire?

Entire is a developer platform designed for a world where AI agents write most of the code. Thomas Dohmke left GitHub in August 2025 and launched Entire in February 2026 with a $60 million seed round led by Felicis, at a $300 million valuation. Other investors include Madrona, Basis Set Ventures and M12, Microsoft's venture fund.

The goal is to build the next developer platform, one where agents and humans collaborate and ship together. Entire does not train models or build its own agents. Instead, it integrates with the agents developers already use, such as Claude Code, Cursor and Codex. The company is fully remote and has around 50 employees.

Entire got here in steps. In February it released Checkpoints, an open source tool that records agent sessions alongside your Git history. In July it launched a distributed Git network that copies existing GitHub repositories to servers in different regions, so agents can download and update code close to where they run.

What launched this week

On 26 October, Entire introduced what it calls a Git forge: a service to host code, review changes and run builds. That is the core of what GitHub does, and Entire positions itself as an alternative. The launch came two days before GitHub Universe, GitHub's annual developer conference. In the launch post, Dohmke describes the platform as built on three foundations: Trails, distributed Git hosting and Continuous Memory.

  • Trails replaces the pull request. A pull request (PR) is the standard way developers propose and review a change. Entire argues that agents now write code faster than people can review it, so the PR has become a log that only the agent reads. A Trail instead puts scores at the top of the page. On every push, several review agents run evals, automated checks that score the change on criteria such as security, sloppy code and whether a human needs to look. Teams can set Gates, rules that block a merge until findings are resolved, a human has approved and tests pass.
  • Distributed Git hosting. Repositories can now be hosted directly on Entire and replicated across regions in the US, Europe, Australia and India, with the UK added as a primary region this week. That helps teams that need to keep code in a specific country. Entire claims pushes up to 89 times faster than GitHub and 6 times faster than Cursor Origin, measured with ForgeMark, a benchmark it has published as open source.
  • Continuous Memory. Agents usually forget everything when a session ends. Entire's answer is two open source tools that run locally on your machine. entire-graph builds a map of how your code fits together and, in Entire's testing, cut token use by up to 71%. Tokens are the units AI models process and charge for, so fewer tokens means lower cost. entire-brain indexes past agent sessions so an agent can recall why earlier changes were made.
  • Build and test pipelines. Through partners Buildkite and Depot, Entire offers continuous integration (CI), the automated building and testing of every change. Both offer migration tools for teams coming from GitHub Actions.

Notably, Entire does not force a move. Teams can keep GitHub as their source of truth, mirror their repositories and even use Trails on top of their existing pull requests.

Why this matters for our community

The bigger story is not one startup. It is that the tools around our code are being redesigned for agents. Entire is not alone: Cursor launched its own code hosting service, Origin, in August. Anthropic has described building its own source control system after its daily merged changes grew from about 2,500 in May to more than 30,000 in September. GitHub reported 7.38 billion commits in September alone, a sign of how much agent activity it already carries.

For our community, this raises practical questions:

  • Review is the new bottleneck. Many of you already run several agents in parallel. When code is cheap to write, the scarce skill is judging it. How teams design review, testing and guardrails deserves a place at every chapter meetup.
  • Context is becoming an asset. Features like Continuous Memory treat the reasoning behind a change as something to keep and share. That matters for anyone building agentic workflows, where agents plan and carry out multi-step tasks on their own.
  • Choice benefits builders. More competition between developer platforms leads to more experimentation, and our community gains when new ideas are tested in the open.

Our take and open questions

The direction makes sense. Pull requests were designed for humans reviewing other humans' work, and that model is under pressure. Still, a first glance leaves us with questions to answer before we form a real opinion:

  • Ecosystem. GitHub has years of integrations, community and open source momentum behind it. That is hard to replicate, and the fact that Entire builds around GitHub through mirroring shows it knows this.
  • Trust in agent reviews. Merging code without reading every line is a big step. How reliable are the eval scores, and how does a team decide when a human must review?
  • Benchmarks. Publishing ForgeMark as open source is a good move, because anyone can rerun it. We would still like to see results from real teams under real workloads.
  • Cost. Running review agents on every push uses compute. How pricing works out at scale is still unclear.

In short, Entire is a bold, well-funded bet on where software development is heading, and one worth following closely.

What's next

Entire says it can only build this platform together with the community, and is collecting feedback on its Discord. We like that spirit, and we want to hear from you too. Have you tried Entire, Trails or Checkpoints with your own agents? What worked and what did not? Share your experience in the community channels, or bring it to your next chapter meetup or AgentCon stop. Want to present your findings? Get in touch. Hands-on sessions from community members are exactly what we love to put on stage.

We will publish a hands-on follow-up once we have built something real with it.

Sources

Written by Stephen Simon

Keep reading

Say hello to n8n, our newest partner!

n8n, the fast-growing workflow automation platform from Berlin, is now a partner of the Global AI Community.

Read the post →
See all blog posts