Ex-GitHub CEO Nat Friedman Launches Entire: Git for AI Agents

Nat Friedman, ex-GitHub CEO, just launched Entire — a platform built for AI agents in software workflows. Backed by a $60M seed at a $300M valuation, it’s open source and focused on making agentic coding observable, auditable, and easier to manage.

Checkpoints: The Missing Link in AI Dev

Entire's killer feature? Checkpoints. When AI agents like Claude Code or Gemini CLI generate code you commit to Git, Checkpoints capture the full session context as first-class Git data:

• Complete transcripts and prompts
• Files touched/modified
• Token usage and costs
• Tool calls and decisions
• Agent reasoning traces

This isn't just logging – it's a Git-native audit trail for AI-assisted work. No more black-box commits.

A $60M Bet on Agentic Infrastructure

The launch sparked debate: some call it incremental, others see a new category of AI-native developer infrastructure. Entire’s bigger pitch is a GitHub-like layer where humans and agents can collaborate with full traceability.

Investors like Felicis, Madrona, and Microsoft’s M12 are backing the bet that trustworthy agent workflows will become core development infrastructure.

Practical Takeaways for Builders

From my workbench at DefendreSolutions:

• Observability first: As we deploy agents in prod, tracing decisions is non-negotiable.
• Open-source hook: Install the CLI today – entire.io – and future-proof your Git.
• Valuation signal: A $300M early valuation signals major confidence in AI developer tooling.
• Builder takeaway: Teams that prioritize auditability and reproducibility will ship safer agent workflows faster.

Watch this space – Entire could be the Git of the agent era.