Former GitHub CEO launches 'Entire' to serve AI-driven 'vibe coding' workflows

Thomas Dohmke, the former chief executive of GitHub, has launched Entire, a new distributed Git hosting network aimed at supporting AI-driven development workflows and the surge of agentic — or “vibe” — coding, the company and press coverage say.

Entire is available in a preview that mirrors public and private GitHub repositories into regional servers in the US, EU and Australia, allowing AI agents to clone, pull and push against an alternate network rather than GitHub itself, according to reporting and Entire’s announcements.

Why this matters: the company frames Entire as infrastructure built for a future where swarms of automated assistants, not just human developers, generate heavy, concurrent Git traffic. If those agent-driven workflows materialise at scale, they will change how version control is provisioned and priced — and test the limits of existing centralized platforms.

How Entire works and the performance claims

Entire’s preview lets developers “mirror an existing GitHub repository onto Entire in one step,” which keeps the canonical code on GitHub while routing agent read and write traffic to regional Entire mirrors, ZDNet reports. Entire also ships a command-line client that the company says can record AI-agent sessions alongside commits — capturing prompts, agent responses, file diffs and contextual metadata — to create an auditable trail for agent activity, The Register writes.

The startup publishes aggressive throughput figures: handling up to 2.1 million pushes per hour and 570,000 clone operations per hour, numbers reported by The Register and presented by Entire in its launch materials. Entire’s coverage contrasts those figures with metrics quoted for Cursor Origin, another newcomer, which the Register cites as claiming roughly 81,000 pushes per hour and 296,000 clones per hour.

ZDNet frames Entire as a distributed network built to “offload heavy, concurrent read traffic so agents can build without rate limits,” and says the system is compatible with several coding agents and models including Codex, Claude Code, Cursor and GitHub Copilot.

How Entire positions itself against GitHub and rivals

Dohmke’s pitch is explicitly reactionary: Entire targets the bottlenecks platform operators face as automated tooling multiplies repository access and repository churn. The Register notes Entire creates “a parallel universe where AI agents can fumble their way through Entire-hosted code without putting strain on GitHub resources that might be needed for actual deployment.”

That positioning places Entire squarely against GitHub’s more centralized model and alongside specialised entrants such as Cursor Origin. Entire’s differentiator, according to coverage, is regional, mirrored hosting and agent-session tracing — features marketed to enterprises worried about rate-limits, latency and auditability for automated developers.

But the launch materials and press pieces leave open important questions. The throughput figures cited are company benchmarks reported by the press; independent third-party audits or customer performance data are not available in the coverage so far. Observers also note that mirroring repositories introduces operational complexity — including sync consistency, access-control mapping and storage costs — that will matter when teams move from experiments to production.

Hazardously, the only public scepticism recorded in initial coverage is implicit: reporters stress Entire’s claims and the problem it addresses, but do not yet produce named external critics or independent tests of the metrics. That gap will matter because vendors routinely publish optimistic performance numbers during previews.

Entire’s timing answers a practical question: as more teams adopt agentic workflows, platform operators face both surplus read-loads and new compliance questions about what AI agents did to code. Dohmke — who left GitHub in 2023 — says the market needs Git infrastructure rethought for these demands, and the preview positions Entire as an early option for companies experimenting with large-scale agents.

Closing: watch for the next concrete signals — enterprise pilot announcements, independent benchmarks, and how GitHub responds on rate-limits and mirrored-read capabilities. Those will determine whether Entire is a niche anti-rate-limit utility or the start of a genuinely new tier of developer infrastructure for the age of automated coding.