How to read this screen. Usage comes from Copilot and Claude Code; outcomes from GitHub pull requests. Together they show adoption alongside delivery. They do not prove that AI caused a change, and the labels say so.
PRs merged, 30 days
412
14 repositories on the allowlist
observedcountGitHub
PRs with AI involvement
61%
251 PRs whose author had AI activity that day
observedratioGitHub + Copilot + Claude Code
Median cycle time
31 h
−18% vs prior 30 days (38 h)
correlationhoursGitHub
Copilot lines added
48.2K
Accepted suggestions, not merged lines
observedcountGitHub Copilot
Cycle time, AI-involved vs other PRs
correlationMedians, opened to merged, in hours. The groups differ in PR size and author seniority, so this is a correlation, not an effect size.
Outside engineering
unavailableSupport, Sales and Finance have no system of record connected. Upload a quarterly outcomes CSV (tickets closed, deals, close time) to move these teams from unavailable to estimated.