Skip to main content

AI-assisted Commits

Commits tagged as AI-authored, alongside the share against total commits in the period.

AI-assisted Commits counts commits Leanmote tags as AI-authored, alongside the share against total commits in the period. It's a finer-grained version of AI-assisted PRs that captures the actual code-authoring touch points.

What it measures

The number of commits in the period where Leanmote's AI commit-tagging logic flags the commit as AI-authored, plus the share against total commits in the same period.

How Leanmote tags an AI commit

A commit is tagged as AI-assisted when either of the following is true:

  • The commit's author had recorded AI activity on the same calendar day the commit was authored, AND the AI activity is associated with that author's identifier.

  • The commit message contains a Co-authored-by: trailer matching a known AI identity (such as github-copilot, copilot[bot], [email protected], claude, cursor-ai, cursor[bot], codeium, or tabnine).

The same tagging logic powers AI-assisted PRs — they should always trend together.

How to interpret it

  • Compare to the total commit count. AI-assisted Commits ÷ Total Commits gives a daily commit-share that's often more responsive than the PR-level metric.

  • High commit share with low PR share — AI is touching many small commits but not always making it into merged PRs.

  • Low commit share with high PR share — AI tools are used to start work, but most commits in the PR were non-AI iterations.

  • Sudden jumps — typically caused by Co-authored-by trailers being switched on or off in the team's commit conventions.

What to do about it

  • Use this as a sanity check on AI-assisted PRs.

  • If commit-level AI usage is high but PR-level adoption is low, investigate whether AI work is hitting friction in review.

Related metrics

  • AI-assisted PRs

  • AI Lines of Code

  • AI Intensity

Did this answer your question?