AI adoption,
measured from the work.

Five levels for every developer, each with its next step.

See the full report

Trusted by engineering teams

Kiwi.comFinshapePartners BankaPurple TechnologyGreysonItrinityESETFTMOKeggSecond Foundation
NYour AI adoption read-outJun 8 to Aug 2 · 8 weeks
  • Agent fluency4.4 of 5, org median 3.04.4↑ +0.5 in 8 weeksKeep AI in the daily loop, not only on the tickets that look hardAI-active 7 of 8 weeks
  • Delegation depth3.9 of 5, org median 2.13.9↑ +0.9 in 8 weeksHand agents whole tasks, not snippets, on every agent dayDeep delegation on 21 of 28 agent days
  • Roadmap alignment3.5 of 5, org median 1.73.5↓ −0.3 in 8 weeksPoint more AI hours at roadmap objectives72% of work credited to roadmap objectives
  • Context leverageStart here0.8 of 5, org median 2.30.8↓ −0.1 in 8 weeksAdd a why and an outcome to every prompt and ticket3 of 12 artifacts state a why and outcome
  • Agentic autonomy0.5 of 5, org median 1.20.5↑ +0.4 in 8 weeksOne agent-owned task a week, start to finish2% of work owned end to end by agents
Org median, 60 developersScored from commits, pull requests and tickets. No survey.
01Across the org

24 developers have never let an agent finish a task.

One bar per level, across all 60 developers. Each row ends in the one thing to do about it.

Not started(below 1)Starting(1 to 2.5)Fluent(2.5 to 4)Leading(4 and up)
  • Agent fluency6142614Most people already use it daily. Get the last 6 started
  • Delegation depth825189Pair the 9 leading developers with the 25 still handing over snippets
  • Roadmap alignment1326156Link tickets to objectives, so AI hours land on the roadmap
  • Context leverage1023207Add a why and outcome field to the ticket template
  • Agentic autonomy242112324 developers have never let an agent finish a task. Start there
02The payoff

Adoption shows up as capacity you did not hire.

The same company, measured before and after AI. Raise the levels above and these three numbers move.

+56engineers of capacity, on a team of 60
  • On the payroll60
  • Delivering like116
$2,230AI cost per engineer of capacity, a month
  • Engineer of capacity, from AI$2,230
  • Team member, all in$8,000
48.6%of that capacity went to roadmap work
Roadmap 48.6%Everything else 51.4%

Adoption is the coding part of the AI transformation report, and one stage of the engineering loop. The capacity is priced in AI ROI, and the bill behind it in token spend.

03How it works

No survey. No leaderboard.

The five levels are never added up. An average hides the lowest row, and the lowest row is the one worth working on.

01

Connect

Read-only access to your repositories, tracker and AI tools.

  • Repositories
  • Tracker
  • AI tools
NRead-only
02

Score

Five levels per developer, read from the work every night.

  • Agent fluency4.4
  • Delegation depth3.9
  • Roadmap alignment3.5
  • Context leverage0.8
  • Agentic autonomy0.5
03

Next step

Every developer sees their own read-out and the one move to make.

Start here · Context leverageYou 0.8Org median 2.3→ Add a why and an outcome to every prompt and ticket
→Start here

See how your team
really uses AI.

Read-only setup, under an hour. The first read-out is built from history you already have.