OpenAI has brought on board the founders of Git AI, an open-source tool designed to monitor AI-generated code and evaluate how well coding agents perform, according to an announcement made late Friday. Aidan Cunniffe and Sasha Varlamov, who created the tool, will become part of OpenAI's Codex team, where they'll focus on providing companies with better metrics about coding agent effectiveness and the financial returns those tools deliver. The move comes less than a year after Cunniffe left his job at Atlassian to transform Git AI from a side project into a standalone commercial venture.
Git AI operates as a Git extension that connects every line of AI-written code to the specific agent, model, and prompts that created it, maintaining that context even after code has been committed, merged, or rebased. The platform helps development teams understand how much AI-generated code actually makes it into production, how frequently it gets revised, and where computational resources and tokens are being consumed. That information can then be used to evaluate different coding agents and models and determine whether the spending on these tools produces valuable results. The tool already works with a wide array of coding agents, including OpenAI Codex, Anthropic's Claude Code, Cursor, Google's Gemini CLI, and background agents such as Codex Cloud, Claude Web, Cursor Agent, and Devin.
Cunniffe wrote in a Friday blog post that the partnership reflects a common focus on measuring how AI coding tools perform and what value they generate. "OpenAI shares our belief that users should have the data to compare model performance and understand the ROI of every token they spend," he stated. Thibault Sottiaux, a technical staff member at OpenAI working on ChatGPT and Codex, confirmed on X that the company plans to leverage Git AI's technology to help businesses grasp Codex's impact. "We'll make it easier for businesses to see where Codex is making a difference when working through problems for individuals and teams," Sottiaux wrote.
The acquisition reflects OpenAI's growing emphasis on helping enterprise customers quantify the business value of AI coding assistants, a question that's become increasingly urgent as companies pour resources into these tools without clear metrics on their effectiveness. Git AI began as a weekend project in summer 2025 when Cunniffe and Varlamov, while still at Atlassian, tried to answer what seemed like a straightforward question: how much of their code was AI-written, and what happened to that code afterward? By November's first major release, the project had expanded well beyond its initial scope, leading Cunniffe to leave Atlassian in January 2026 and build a commercial business around the open-source foundation. While the fate of Git AI as a standalone commercial entity remains unclear with both founders now joining OpenAI, Cunniffe says the company will keep supporting the open-source project. "We'll keep investing in open source, while also giving enterprises the data they need to build effective software factories," he noted.
How Git AI's cross-model independence survives under OpenAI's ownership will be worth watching, particularly as the tool currently tracks performance across competing platforms including Anthropic and Google. The integration could give OpenAI a significant advantage in convincing enterprise customers that Codex delivers measurable value, especially if the analytics infrastructure favors OpenAI's own products over rivals. For businesses already struggling to justify AI coding tool budgets, having vendor-provided metrics rather than neutral third-party measurement may complicate rather than clarify purchasing decisions.