Former GitHub CEO Returns with $60M Seed to Rebuild Coding for the Agent Era

In what is being described as the largest seed funding round in the history of developer tools, Thomas Dohmke, the former CEO of GitHub, has officially unveiled his new venture, Entire. Emerging from stealth with a $60 million injection of capital and a $300 million valuation, Entire aims to dismantle and reconstruct the software development lifecycle (SDLC) to accommodate the explosive rise of AI coding agents.

The round was led by Felicis, with significant participation from Microsoft’s venture fund M12, Madrona, and Basis Set Ventures. The startup has also attracted a roster of high-profile angel investors, including Y Combinator CEO Garry Tan, Datadog CEO Olivier Pomel, and Yahoo co-founder Jerry Yang, signaling strong industry confidence in Dohmke’s thesis: the tools that built the open-source era are insufficient for the age of AI generation.

The End of "Craft" Software Development

For decades, software engineering has been treated as a digital craft—human developers writing logic line-by-line, committing changes to version control systems like Git, and manually reviewing pull requests. Dohmke argues that this model is rapidly becoming obsolete as AI agents begin to generate code at a volume and velocity that human workflows cannot sustain.

"We are living through an agent boom, and now massive volumes of code are being generated faster than any human could reasonably understand," Dohmke stated at the launch. "The truth is, our manual system of software production—from issues, to git repositories, to pull requests, to deployment—was never designed for the era of AI in the first place."

Entire’s philosophy draws a direct parallel to the Industrial Revolution. Just as the automotive industry shifted from artisanal workshops to moving assembly lines to achieve scale, software development must transition from human-centric tooling to an infrastructure designed for "manufacturing" code via agents.

Introducing "Checkpoints": Versioning Logic, Not Just Code

The immediate problem Entire seeks to solve is the "black box" nature of AI-generated code. When an agent like Anthropic’s Claude or OpenAI’s models generates a script, the reasoning, prompts, and context behind that code are typically lost the moment the file is saved. This loss of context creates what the industry has termed "AI slop"—code that works but is unmaintainable because its intent is opaque.

Entire’s first public offering is Checkpoints, an open-source command-line interface (CLI) tool. Unlike standard Git commits that only save the resulting code, Checkpoints captures the entire "session context" of the AI agent. This includes:

• The original prompts given to the agent.
• The agent's "chain of thought" or reasoning process.
• The specific tools and constraints applied during generation.

By storing this metadata alongside the code in a Git-compatible database, developers can "replay" the creation process, allowing for true auditability and easier debugging of agent-generated software.

A New Infrastructure Stack

While Checkpoints is the entry point, Entire’s ambition extends to building a full-stack platform that acts as the nervous system for AI development. The company is developing a three-layer architecture designed to replace or augment existing CI/CD workflows:

1. Git-Compatible Database: A storage layer that unifies code with its generative context, ensuring that "why" a change was made is as accessible as "what" was changed.
2. Semantic Reasoning Layer: A control plane that allows multiple AI agents to coordinate, effectively acting as project managers that can understand the intent behind the codebases they are modifying.
3. AI-Native Interface: A user experience designed not just for typing code, but for orchestrating fleets of agents, reviewing high-level specifications, and intervening only when necessary.

Investment Overview

The $60 million seed round is an anomaly in the current venture capital climate, where seed rounds typically range from $1 million to $5 million. The size of the raise reflects both the capital intensity of building foundational infrastructure and the track record of the founder. Dohmke, who led GitHub during the launch and scaling of Copilot, is uniquely positioned to understand the limitations of the current ecosystem.

Funding Round Details

The "Drift" Dilemma

One of the critical technical challenges Entire addresses is "drift." As AI agents iterate on code, they can inadvertently drift away from the original project specifications or introduce subtle bugs that accumulate over time. Traditional code review processes—where a human reads every line of a diff—are becoming bottlenecks.

Entire’s platform proposes a shift from reviewing code to reviewing specifications and outcomes. By capturing the intent at the source, the platform aims to allow humans to govern the process of software generation rather than inspecting every output. This aligns with the broader industry trend toward "Agentic" workflows, where humans move "up the stack" to become architects and supervisors of autonomous coding bots.

Market Implications

The launch of Entire poses a potential challenge to incumbent platforms like GitHub and GitLab. While these platforms have integrated AI assistants (like Copilot and Duo), their underlying architecture remains rooted in Linus Torvalds’ 2005 vision of Git—a tool for human collaboration.

Dohmke’s bet is that retrofitting AI onto these platforms is insufficient. If his vision holds true, the next generation of software won't just be written by AI; it will be managed, versioned, and deployed by infrastructure that treats human code as the exception, not the rule. With the release of Checkpoints, developers can now begin testing this hypothesis, effectively version-controlling their AI's "thoughts" for the first time.