AI Tool Adoption Tracking
Track real AI coding assistant usage across GitHub Copilot, Cursor, and Claude Code — not just licenses, but actual activity, acceptance rates, and delivery impact.
The AI Analytics dashboard (/ai-analytics) gives engineering leaders a single view of how your team actually uses AI coding tools — who is adopting them, which tools are driving value, and whether AI assistance is moving your cycle time.
Koalr tracks GitHub Copilot, Cursor, and Claude Code (via the Anthropic integration). This is Koalr's strongest competitive differentiator: no other engineering metrics platform surfaces real AI tool usage data.
What's tracked
GitHub Copilot
| Metric | Description |
|---|---|
| Adoption rate | % of seats with ≥1 accepted suggestion in the last 28 days |
| Acceptance rate | Accepted suggestions ÷ total suggestions shown |
| Lines accepted | Total AI-generated lines accepted per day |
| Est. hours saved | Accepted lines converted to estimated engineering hours |
| Top editors | IDE distribution for Copilot usage |
| PR impact | Cycle time correlation: Copilot users vs. non-users |
Cursor
| Metric | Description |
|---|---|
| Daily Active Users (DAU) | Developers who made ≥1 AI request in a given day |
| Total requests | Aggregate AI requests per day (30-day trend) |
| Model breakdown | Distribution across claude-3-5-sonnet, gpt-4o, cursor-small, etc. |
| Spend | Total and per-member cost (Admin view only) |
Combined view
The top of the AI Analytics page shows a 30-day combined DAU chart overlaying all connected AI tools — so you can see total AI coding activity across your team in one line, regardless of which tools individuals prefer.
AI-authored commit detection
Separately from usage API data, Koalr scans every pull request for Co-authored-by Git trailers to detect AI-generated commits at source. This works even for tools that don't expose a usage API. Detected patterns include:
Co-authored-by: GitHub Copilot <[email protected]>Co-authored-by: Claude Code <[email protected]>Co-authored-by: Cursor <[email protected]>Co-authored-by: Codeium <[email protected]>Co-authored-by: CodeWhisperer <[email protected]>
When detected, Koalr surfaces an informational signal in the GitHub Check Run: "AI-authored code detected." This is not a risk penalty — it's a visibility signal. See Deploy Risk: GitHub Check Run for details.
Connecting AI tools
| Tool | Integration path |
|---|---|
| GitHub Copilot | Integrations → AI Tools → GitHub Copilot |
| Cursor | Integrations → AI Tools → Cursor |
| Claude Code (Anthropic) | Integrations → AI Tools → Anthropic |
Each tool shows a connect flow with an IntegrationGate prompt if not yet configured. You only see data for tools you've connected.
Interpreting cycle time correlation
The PR Impact section shows average cycle time for developers who actively use AI tools vs. those who don't, over the selected date range.
Factors that affect this comparison:
- Developers vary in seniority and PR complexity — this is a correlation, not a controlled experiment
- Look for trends over time (is the gap growing?) rather than a single snapshot
- Teams with fewer than 5 developers in either cohort will show "insufficient data"
Visibility and privacy
- Aggregate metrics are visible to all dashboard users
- Per-member spend and individual usage counts are visible to Admins only
- No code content, prompts, or completions are ever accessed — Koalr reads usage metadata only
- Visibility follows your org's global setting (Full Transparency vs. Role-Based)