Ex-GitHub CEO Thomas Dohmke Launches Entire: A $60M Platform to Solve AI Code's Context Crisis

When an AI coding agent generates a 500-line diff, the reasoning behind it — the prompts, tool calls, and decisions — disappears the moment the terminal session closes. This is the problem Thomas Dohmke, the former CEO of GitHub, set out to solve with his new startup, Entire.

Launched on February 10, 2026, Entire is a developer platform built for the age of AI coding agents, backed by a record-breaking $60 million seed round at a $300 million valuation . Its first product is an open-source CLI called Checkpoints, designed to make the 'why' as permanent and searchable as the 'what' in your Git history .

What Problem Does Entire Solve for AI Coding Agents?

Entire addresses what Dohmke calls the next bottleneck in software development: the reviewability, auditability, and transparency of AI-generated code .

When human developers write code, their intent can often be inferred from commit messages, code comments, and design documents. But when an AI coding agent writes code, its session context — the prompts, reasoning steps, tool calls, and intermediate decisions — is ephemeral. It vanishes when the terminal session ends . Traditional Git captures the 'what' (code diffs and file states) but not the 'why' (the reasoning process) .

Entire means to make that agent session context a first-class, versioned, and searchable part of the repository, creating what some have described as a "black box flight recorder" for AI-generated code .

Key Features: Checkpoints and Beyond

The first product from Entire is Checkpoints, an open-source CLI (MIT licensed) that hooks directly into Git workflows . Its core functionality is straightforward:

• Automatic Context Capture: When a developer or agent runs

  git commit

  , Entire automatically captures the full agent session — including prompts, responses, tool calls, files touched, and token usage — and attaches it to the commit as a permanent checkpoint .
• Searchable Provenance: Developers can later query "why did this code change" and see the full transcript that led to any specific diff, enabling much deeper code review and debugging .
• Multi-Agent Support: Checkpoints works with a range of popular AI coding agents, including Anthropic's Claude Code, OpenAI's Codex, Google's Gemini CLI, OpenCode, and others .
• Clean Git History: Session metadata is stored on a dedicated entire/checkpoints/v1 branch, keeping the user's active branch clean and preserving a traditional Git workflow .
• Upcoming Context Graph: Beyond the CLI, Entire plans to ship a broader 'semantic reasoning layer' called the context graph, designed to help developers navigate the decision trees of multiple agents across the entire repository .

The Entire.io website also mentions a repo mirroring feature designed for "maximum agent throughput," and the GitHub organization (entireio) hosts the open-source CLI code .

Benchmarks and Performance

The reporting around quantitative benchmarks is limited. Available sources describe performance claims in qualitative terms — for example, that Checkpoints makes it "instantly" searchable to recover the context behind any line of code . No standardized quantitative benchmark numbers (e.g., latency, throughput, code review time reduction) were found in the credible sources reviewed. This appears to be a gap in public disclosure as of the launch period.

Funding Background

Entire launched with what lead investor Felicis called the largest seed round ever for a developer tools company .

Detail	Fact
Amount raised	$60 million seed round
Valuation	$300 million
Lead investor	Felicis
Other investors	Madrona, Microsoft's M12, Basis Set
Notable angels	Gergely Orosz, Harry Stebbings, Jerry Yang, Olivier Pomel

How Entire Fits into the Broader Trend

Entire is one of several high-profile moves signaling that the developer tooling stack is being fundamentally rethought for an AI-agent-heavy world:

• Open-source and independent. Entire is positioning itself as an open, independent platform — not tied to any single AI model vendor. Dohmke has emphasized that openness should sit at the heart of the next developer platform .
• Git-native, not Git-replacement. Entire extends Git rather than replacing it. It treats agent context as first-class Git metadata, a different approach from fully decentralized alternatives like Radicle or GitButler's reimagined Git client .
• Parallel moves by other GitHub founders. Scott Chacon (co-founder of GitHub) raised $17M from a16z for GitButler, a reimagined Git client for the AI era. Both Dohmke (Entire) and Chacon (GitButler) are betting that the existing Git workflow is insufficient for agent-generated code, but they take different architectural approaches — Entire focuses on context capture and provenance, while GitButler focuses on a new Git client UX .
• Human-agent collaboration as a first principle. Entire's design philosophy is that humans and AI agents will co-author code, and version control must capture the human's intent and the agent's reasoning together — not just the file diffs . This is a broader industry shift away from human-only Git workflows toward "agent-native" DevOps infrastructure.