Thought Leadership
GitHub appears to be reaching its limits as AI coding agents generate an unprecedented volume of repository activity. Former GitHub CEO Thomas Dohmke believes he has a better approach.
After leading GitHub for years, Thomas Dohmke has unveiled Entire, a new Git hosting platform built around a simple premise: traditional services like GitHub were never designed for a world where autonomous AI agents write code and continuously modify repositories.
A Name With a Message
The company name Entire is intended to double as a slogan. It describes the platform as an “entirely new” Git hosting network built on Git, the version control system that has been the industry standard for more than two decades.
In a blog post, Dohmke argues that Git’s future is not in question. Given its massive adoption, the real challenge is how Git repository hosting must evolve in an era where AI agents generate the majority of new code.
Git itself remains firmly entrenched. An estimated 94 percent of developers rely on the technology, making it by far the world’s most widely used version control system. The bottleneck, according to Dohmke, is GitHub — the Microsoft-owned hosting platform — which is increasingly strained by the surge of requests generated by AI coding agents.
Returning to Git’s Decentralized Roots
Dohmke openly acknowledges that Entire’s core idea—decentralization—has always been part of Git’s original design. The company is therefore addressing fewer shortcomings in Git itself than in the way major hosting platforms have implemented it.
Git was built from the outset as a decentralized system. Every cloned repository contains the complete project history, allowing software to be distributed across multiple servers instead of relying on a single central platform. In practice, however, major hosting providers have encouraged developers to work within centralized infrastructures. That model worked well for years but is now reaching its limits as AI agents dramatically increase repository activity.
“We believe Git hosting must return to its original promise: a truly decentralized network, not a system where the world’s software is stored in one place. That’s exactly what we’ve built.”
Thomas Dohmke
How Entire.io Works
Entire.io is currently available as a limited test with a waitlist for users in the United States, the European Union, and Australia.
Developers can mirror both public and private GitHub repositories, creating a parallel environment where AI agents can work without placing additional load on GitHub’s production infrastructure. Alternatively, teams can host branches directly on Entire.
The platform also includes the Entire CLI, a command-line tool tightly integrated with Git. It records AI agent sessions alongside standard Git commits. According to the company, it captures prompts, agent responses, file changes, and additional context, making it possible to understand not only what changed but also why an AI agent made a particular decision.
That also hints at Entire’s business model: Git hosting combined with AI decision logging and auditing, with the potential to build additional services around the resulting data.
Performance Claims Compared With Cursor Origin
Entire says its decentralized network is designed for high scalability, low latency, regional data control, and high availability, enabling it to handle the massive number of concurrent operations generated by AI agents.
To illustrate its performance, the company compares Entire with Cursor Origin, the AI agent-focused Git platform introduced by Cursor last month.
According to Entire, its network can process 2.1 million Git pushes per hour and 570,000 repository clones per hour. By comparison, Entire claims Cursor Origin handles 81,000 pushes and 296,000 clones per hour.
Open Source Plans Ahead
Over the coming months, Entire plans to release its network as open source, allowing organizations to self-host the platform.
In a post on X, Dohmke said the company intends to continue developing the entire technology stack with the goal of creating an open, decentralized, and independent developer ecosystem designed for both AI agents and human developers.
(lb)
