Entire, a developer platform founded by former GitHub CEO Thomas Dohmke, is now live. The company positions the service as a direct alternative to GitHub, specifically engineered to handle the rapid pace of AI-generated code. The platform relies on a decentralized Git network announced in July, which the company claims can withstand the infrastructure strain caused by surging AI coding activity.
Replacing Pull Requests with Trails
The most significant structural change in the Entire git platform is the removal of traditional pull requests. Instead, developers use a feature called Trails. According to the company, Trails function as a heads-up display that provides customized evaluations scored by AI agents on every code push. This approach aims to streamline the review process for teams merging changes as quickly as agents create them.
Thomas Dohmke stated that the platform treats the idea, agent trace, and outcome of a developer session as first-class artifacts for review alongside the code itself. The system uses multiple agents to generate evaluations that verify the quality and success of AI-generated code. This allows developers to merge code without manually reviewing every line in a diff, provided all evaluations are satisfied.
All data within the platform is accessible through the Entire web interface and an open-source command-line interface. This setup allows coding agents to address findings in a continuous loop while human developers validate the progress. The agents running these processes are customizable, and each trail displays how agents are spending tokens.
Decentralized Hosting and Continuous Memory
Entire claims its decentralized network offers Git hosting speeds up to 89 times faster than GitHub and six times faster than Cursor origin. The platform enables developers to create, migrate, and host private repositories across multiple global locations. Initial deployment regions include the United States, India, Australia, and the European Union.
Another key feature is Continuous Memory, which stores developer session logs next to the code in the repository. This data source allows agents to search for answers, which the company says reduces token usage by up to 71 percent. The memory system consists of a codebase graph and an agentic brain that uses previous sessions and prompts as lessons to prevent agents from repeating mistakes.
The launch comes as AI tools increasingly automate coding tasks, creating new demands on version control infrastructure. Entire aims to address these demands by redesigning features that developers have become accustomed to, focusing on speed and agent integration.
Source: ZDNet
