Entire, a Git-compatible distributed source control server led by Thomas Dohmke, launches a GitHub rival built for AI agents, promising up to 89x faster code pushes than GitHub and six times faster than Cursor Origin, with a new Continuous Memory feature that lets AI agents search past coding sessions for context and enables persistent knowledge from prior work.

Continuous Memory includes a codebase map and two CLI plugins, entire-graph and entire-brain, to map code and attach agent sessions for durable, searchable memories visible to reviewers in Trails.

Trails replaces traditional pull requests with an AI-assisted evaluation dashboard that scores code changes from multiple agents, while allowing human review or automated tests to gate merging.

Preview access is free at launch, with no disclosed pricing details at this time.

Entire previously raised $60 million in February and plans to broaden offerings to larger enterprises and broader developer communities.

The product is available to try today for all developers.

The company invites community feedback via Discord to help shape its forge of the future.

The launch aligns with a broader shift toward AI-assisted software development and evolving code-review processes, mirroring industry trends highlighted at GitHub-related events.

Thomas Dohmke emphasizes maintaining bridges with GitHub and Microsoft, aiming to attract large enterprise customers and foster ongoing relationships within the ecosystem.

Microsoft announced a voluntary exit offer affecting about 60,000 employees, a development noted in the broader tech context surrounding the launch.

Entire enables AI agents to propose and fix issues automatically, while developers retain the ability to review outcomes and merge when checks pass.

Partnerships with Buildkite and Depot integrate build and test pipelines with AI-enabled workflows, offering migrations away from GitHub Actions and enabling CI/CD tied to Trails.