Know exactly how much
code AI writes.

Not estimates. Not surveys. Real attribution from every commit. Tandemu measures AI vs manual code using Co-Authored-By tags — the same standard GitHub uses.

See the Dashboard

AI Attribution

Last 30 days

73%

AI Generated

27%

Manual

Per Developer

Sarah Chen

2,340 lines

78%

James Park

1,870 lines

65%

Alex Rivera

2,100 lines

72%

Mia Thompson

1,420 lines

54%

How it works

Claude Code automatically adds Co-Authored-By: Claude to every commit it helps write. Tandemu reads these tags during /finish and attributes lines to AI or manual — per file, per commit.

git log --oneline

a1b2c3d

feat: add webhook retry with exponential backoff

Co-Authored-By: Claude <noreply@anthropic.com>

e4f5g6h

fix: handle edge case in date parsing

(manual commit — no AI tag)

i7j8k9l

refactor: extract auth middleware into guard

Co-Authored-By: Claude <noreply@anthropic.com>

Measure what matters

AI attribution at every level — individual, team, and codebase.

Team AI Ratio

Overall percentage of AI-generated code across the team. Track adoption trends over weeks and months.

Per-developer breakdown

See who's leveraging AI heavily and who might need enablement. Not for judgment — for coaching.

AI effectiveness by file

Which files have high AI contribution that survives? Where does AI-written code get reverted or rewritten?

Investment allocation

How engineering time splits across features, bugs, tech debt, and maintenance — broken down by AI vs manual.

First-party data, not inference

Other tools estimate AI impact from external signals — installed extensions, survey responses, or API call counts. Tandemu measures it directly from the code. Every line attributed at the commit level. No guessing.

Supported tools

Claude Code

CursorSoon

WindsurfSoon

CodexSoon

Know exactly how muchcode AI writes.