Every week, leaders get one narrative report: the numbers that matter, then AI-generated observations with citations to the sessions behind them. This sample uses illustrative data. Reports are always org-level; each person's own report is private to them.
AI work grew 12% this week and fully-loaded cost per merge fell another 8%. Two things need attention: spend is projected to cross budget on day 24, and a flaky staging environment is burning data-engineering sessions. One win to bank: the payments team's migration fix is ready to become a skill.
1,847 sessions, summarized from their session stories. The mix of work:
Four sessions this week hit the same phased-migration dead end before a fifth solved it cleanly. Codiedev flagged the working pattern as a skill candidate; promoting it puts the fix in every future migration session automatically, and would have saved an estimated 11 hours this week alone.
Sessions that end without shipped work rose from 9% to 19% this week and drove the org-wide spike above. The session stories consistently cite a flaky staging environment, not model or prompting problems. One infrastructure fix recovers the whole regression.
Contract-review sessions tripled this month and converged on a consistent redlining approach. Capturing it as shared context would carry the pattern to the two regional counsel teams that haven't adopted it yet.
Current spend projects 9% over plan before month-end. The growth is concentrated in two teams running long investigation sessions against a model mix that costs 2.4× the org median for equivalent work. A model-mix change closes the gap without touching output.
In 20 minutes, we'll show you the record of AI work no other platform can produce, and how it compounds.