THE TOY

Walk vs. Dashboard

You are the engineering manager of a six-pair team, one quarter after it “adopted AI.” In front of you are two views of the same team. The dashboard is real: nothing in it is faked, massaged, or wrong. The walk is also real: six workstations, each with its transcripts open.

At the end of the quarter you choose exactly one intervention. That is the whole game — that, and deciding where to look before you choose. Five minutes.

Walk vs. Dashboard

AI adoption
84%
▲ up from 76%
Throughput
+40%
▲ vs. last quarter
AI-assisted PRs
61%
▲ of all merges
Avg. cycle time
−18%
▲ improving
THROUGHPUT, WEEKLY
  • Pair 1 · MounirSessions/week: 31 · acceptance rate 74%
  • Pair 2 · SanaAI-assisted merges: 9 of 9 · velocity ▲
  • Pair 3 · PetraSessions/week: 44 — highest on the team ▲
  • Pair 4 · InesSessions/week: 12 · acceptance rate 81%
  • Pair 5 · ViktorDaily active ✓ · streak: 41 days ▲
  • Pair 6 · AlbaStory points: 21 ▲ · velocity best on team

Every number above is true. The aggregates are honest. You may choose your intervention from here — a steering committee would.

END OF QUARTER — CHOOSE ONE INTERVENTION

Both views are truthful. Only one shows the work. Hidden in the six transcripts are six things a walker would see and an average cannot.

If you came here directly: this toy is the payoff of a longer walk that starts at a chalk circle — it plays fine on its own, and plays better at the end of the street.