Thomas Dohmke, the former CEO of GitHub, has launched Entire, a developer platform designed to help software teams manage code written by artificial intelligence (AI) agents. The launch puts Dohmke in direct competition with the developer platform he once led, as AI coding tools increasingly change how software is built, tested and reviewed.
Unveiled on October 26, 2026, Entire combines code hosting, automated reviews and persistent memory for AI coding agents. The company’s pitch is straightforward: developers need new tools to review and manage code at the speed AI agents can generate it.
The launch follows Entire’s $60 Mn seed funding round in February, which valued the startup at $300 million. The round was led by Felicis, with participation from Madrona, Microsoft’s venture arm M12, Basis Set and other investors. The latest product launch is not a new funding announcement.
Entire Introduces Trails to Rethink Code Reviews
One of Entire’s main features is Trails, a code-review system designed around AI-generated changes rather than the traditional pull-request process.
In conventional development workflows, engineers submit code changes for colleagues to inspect before they are merged into a project. As AI agents generate more code, reviewing every change manually can become a bottleneck.
Trails aims to address this by running customised evaluations whenever new code is submitted. Multiple AI agents can examine the changes, identify potential problems and work through fixes, while developers monitor progress through Entire’s web interface or command-line tool.
The system is designed to give developers a broader view of the code, the instructions behind it and the results of automated checks. Teams can set evaluation criteria and use the results to help decide when changes are ready to merge.
The approach could reduce the amount of routine manual review required, but automated evaluations do not eliminate the need for human oversight. Teams will still need to establish suitable quality and security standards, particularly for production systems and sensitive software.
Entire Claims Git Hosting Up to 89 Times Faster Than GitHub
Entire is also launching distributed Git hosting, allowing developers to create, migrate and manage private code repositories on its platform.
The company says its infrastructure can handle code pushes up to 89 times faster than GitHub and six times faster than Cursor Origin, a competing code-hosting service. These are company-reported benchmark claims, not independently verified comparisons across every workload.
Developers can host private repositories in the United States, Europe, India and Australia. Entire says it plans to expand its regional infrastructure over time.
The distributed approach is intended to support teams working with large volumes of AI-generated code. Faster code transfers could help in some workflows, although overall development speed will also depend on testing, review, infrastructure and the complexity of the software being built.
Entire is also partnering with Buildkite and Depot to connect code hosting with software build and testing pipelines. Developers can adapt existing GitHub Actions workflows to work with these services, with build results surfaced through Trails. The aim is to reduce delays between generating code and checking whether it works.
Continuous Memory Helps AI Agents Reuse Past Coding Context
The third major feature is Continuous Memory, which allows AI coding agents to search information from previous development sessions stored alongside code in Git repositories.
The system is intended to preserve useful context, including earlier prompts, decisions and coding history, so agents can retrieve relevant information instead of repeatedly starting from scratch.
Entire says its codebase graph and memory system can reduce token usage by up to 71% in its testing. That figure is a company claim, and actual savings will depend on the project, agent and workflow.
Persistent memory could become increasingly valuable as teams use multiple agents across large codebases. It may help preserve decisions and reduce repeated work, although the usefulness of stored context will depend on its accuracy and relevance.
Entire’s first product, Checkpoints, launched in February 2026 as an open-source tool that records AI-agent context alongside Git commits. The new platform extends that idea from tracking agent activity to hosting, reviewing and managing the resulting code.
Thomas Dohmke’s $60 Mn Bet on the Future of Software Development
Dohmke left GitHub in 2025 after nearly four years as its CEO. During his tenure, GitHub Copilot became a prominent AI coding assistant, giving him experience with the shift towards AI-assisted software development.
Entire is not simply building another coding assistant. Instead, it is targeting the infrastructure surrounding AI-generated code: version control, review, testing and the preservation of development context.
The company’s $60 million seed round was led by Felicis, with other institutional investors and technology industry figures backing the venture. Its investors include Garry Tan, CEO of Y Combinator; Olivier Pomel, CEO of Datadog; and Jerry Yang, Yahoo’s co-founder. Entire has said its team includes about 50 people, with experience at companies such as GitHub, Microsoft, Atlassian and DigitalOcean.
The launch comes as developer-tool companies compete to serve teams that increasingly rely on AI agents. Entire plans to expand its offering to open-source maintainers and large enterprises, with public repositories, self-hosted Git nodes and additional integrations among its future priorities.
For developers and businesses, the larger question is whether AI-generated code can be produced, checked and deployed reliably at scale. Entire is betting that the next major opportunity in software development lies not just in generating code faster, but in building the systems needed to verify and manage it.
Its success will depend on whether developers trust its review process, find its hosting advantages meaningful and are willing to move established workflows to a new platform.
