Git AI
1 known investors
Git AI builds a Git extension and observability stack that tracks AI-generated code from prompt to production, linking each line to the agent, model, and prompt that created it. It provides attribution, governance, ROI measurement, and token accounting for software engineering teams and enterprises using AI coding agents.
Also known as git-ai · git-ai-project/git-ai
Investors · 1
Company profile
researched Aug 2026Git AI develops an open source Git extension that records line-level attribution for AI-generated code, linking each line to the coding agent, model, and session that produced it. Coding agents invoke `git ai checkpoint` after each Edit, Write, or Bash tool call; the checkpoints are stored in .git/ai and consolidated on commit into an authorship log attached as a Git note under refs/notes/ai. Because notes are addressed by commit SHA, Git AI rewrites and merges attribution when history-rewriting operations occur, preserving attribution across rebase, merge, merge --squash, cherry-pick, revert, commit --amend, reset, stash/pop, checkout/switch, and pull/push/fetch, typically within 5-100ms of the operation completing. The project states it does not use heuristics, filewatchers, or AI detection: agents explicitly report the lines they wrote.
On top of attribution, the company builds observability, governance, and "harness engineering" tooling. Traces follow AI code through the software development lifecycle, capturing prompts, responses, tool calls, token usage, human overrides, code review comments and requested changes, reworked and churned lines, and production incidents. Dashboards report metrics such as percent AI code merged to production, daily token spend, ratio of lines generated to lines shipped, rework rates, and per-pull-request percent AI, cost, author, agent, and model. Additional capabilities include mining agent sessions to build reusable skills, rules, and context; linking requirements and architecture decisions to the code they produced; and measuring agent "readiness" by area of the codebase.
Products are split between a free, local-first open source CLI that requires no login and works offline, and a teams/enterprise version that adds a secure, redacted prompt store, hosted notes storage, SCM integration (GitHub, GitLab, Bitbucket, Azure DevOps), aggregate statistics across repositories, token-spend attribution to pull requests, and incident-to-session tracing. The teams product can be self-hosted or run in Git AI's cloud, and the CLI can be distributed to developer endpoints via MDM.
Business model
Open core: a free, open source Git extension distributed via an install script for macOS, Linux, and Windows, alongside a paid teams/enterprise offering that adds a secure prompt store, SDLC data joins, cross-repository analytics, and full agent-session observability, available self-hosted or in Git AI's cloud.
The open source CLI is free; revenue comes from the Git AI for Teams/Enterprise product, offered as a hosted cloud service or self-hosted deployment with organization-wide rollout, prompt storage, and cross-repository analytics.
Traction
The open source repository git-ai-project/git-ai has roughly 2,500 stars, 276 forks, and 5,436 commits. Git AI reports adoption by hundreds of engineering teams, including many in the Fortune 100. Integrations cover local agents (Cursor, Claude Code, Copilot, Codex, Windsurf, Gemini CLI, opencode, Continue, Droid, Junie, Rovo Dev, Amp) and background agents (Claude Web, Codex Cloud, Cursor Agent, Devin), plus community IDE plugins.
▸Full profile — market position, technology, go-to-market
Market position
Positions itself as an observability, governance, and harness-engineering layer for AI-assisted software development, describing its category as observability for the "software factory" and its Git Notes format as an open standard for AI code attribution rather than a single-vendor tool.
Attribution is explicit rather than inferred: agents report the exact lines they wrote instead of relying on heuristics, filewatchers, or AI detection. The implementation is Git-native, built on Git Notes as what the company calls an open standard for tracking AI-generated code, is not tied to a single agent vendor, adds no measured overhead because it neither uses Git hooks nor wraps the Git binary, and requires no per-repo setup or workflow changes. Attribution is designed to survive rebases, squashes, merges, stashes, resets, and cherry-picks, and works across worktrees and background agents.
Technology
A Git extension written primarily in Rust that stores line-level AI attribution in Git Notes (refs/notes/ai). It registers pre/post tool-call hooks for supported coding agents, which call `git ai checkpoint` to mark newly written lines as AI- or human-authored; checkpoints live in .git/ai until commit, when they are consolidated into an authorship log note mapping files to agent sessions and line ranges plus JSON metadata (agent, model, human author, messages_url). Attribution diffing and merging handles Git rewrite operations. The tool avoids Git hooks and does not wrap the Git binary, which the company describes as adding zero overhead; it is local-first and works offline. Commands include `git ai stats` (percent AI, generated vs. committed lines, accepted rates, human overrides broken down by tool and model, JSON output), `git ai blame` (a drop-in git blame replacement supporting standard flags), and `git ai install-hooks`. Agent sessions are stored outside Git after scanning and redaction, locally, in Git AI Cloud, or in a self-hosted store. Known gaps include git mv, filter-branch, git replace, web-UI merges, and multi-repo-root Bash attribution.
Go-to-market
Bottom-up distribution through a free open source CLI installed via a one-line curl or PowerShell command and a public GitHub repository, with community touchpoints including Discord, "meet the maintainers" calls, IDE plugins (VS Code, Cursor, Windsurf, Antigravity, Emacs magit), and contributor-added agent support. Enterprise motion runs through "Book Demo" and onboarding calls, MDM-based fleet rollout, and SCM connections for organization-wide analytics.
Software engineering teams and enterprises using AI coding agents, from individual developers using the free local CLI to organizations rolling the extension out to every developer endpoint via MDM; the company says hundreds of engineering teams, including many in the Fortune 100, use it.
Compiled by commissioned research from 8 cited public sources — announcements, filings, and press listed under research sources below.
Key figures
latest reportedCompany-reported or press-reported figures, each dated to when it was claimed — not independently audited.
▸Research sources · 8
primary sources listed
- Git AIusegitai.com · web
8 public sources were cited for this profile; the first-party ones are listed here.
Frequently asked questions
- What does Git AI do?
- Git AI is an open source Git extension and observability platform that attributes AI-generated code from prompt to production.
- Who are Git AI's investors?
- Git AI's investors include Work-Bench.