AI User Engagement Segments classifies AI tool users by how often they engage. It complements the daily-volume views (PRs merged per day with AI Intensity badge, distinct AI users per week) by giving you a per-person engagement profile.
The four segments
Each person's active-day ratio is their active days divided by the calendar days in the period. The cut-offs are inclusive at the bottom:
Active-day ratio | Segment |
50% or more | Power user |
10% up to 50% | Casual user |
Under 10% | Idle user |
Somebody at exactly 50% is a Power user, and somebody at exactly 10% is Casual — the boundaries belong to the higher segment.
New users are counted separately: a person is new when their first AI activity ever falls inside this period. It's computed across all of their tool identities, so somebody who already used one tool and picked up a second this period is not miscounted as new.
"Active" means at least one recorded interaction that day with GitHub Copilot, Claude Code or Cursor.
Other connected AI tools do not reach this panel. Gemini is the case to know about: it is ingested and appears in the AI usage views, but every Gemini row is tagged with the app and the entry point it came from, and these segments only read rows with no such breakdown. Gemini users are invisible here — an empty-looking segment is not proof that nobody is using AI.
Companion daily/weekly metrics
The same dashboard panel surfaces two supporting views:
Distinct AI users per week — count of unique users who used any AI tool at least once per calendar week. Useful for tracking breadth of adoption over time.
PRs merged per day — daily PR merge volume during the period. The badge on this metric shows the estimated AI tooling cost per PR merged: total cost in the period divided by PRs merged.
How to interpret it
Many Power users, few Casual — adoption is concentrated. The team has internal AI champions but the practice hasn't spread.
Many Casual or Idle users, few Power — adoption is shallow. Seats may be paid for but underused.
Lots of New users in a single period — onboarding has just kicked off; expect Casual and Power counts to climb in the next few periods.
Distinct weekly users plateauing — the natural ceiling for the current pool. Further growth requires onboarding more people.
What to do about it
Pair Power users with Idle or Casual users for knowledge transfer.
Audit Idle users with paid seats — reclaim seats or run targeted enablement.
When AI Tool Cost is rising but the count of Power users isn't, you may be paying for seats producing little value.
Related metrics
AI Tool Cost
AI-assisted PRs
AI-assisted Commits
AI Return metrics overview