AI Ops leadership visibility project (Docebo, July 2026) — key constraints established with CITO Nitin Chopra:

• Nitin explicitly REJECTED ROI/dollar-savings framing for AI Ops reporting: "I didn't ask you guys to tell me ROI... if I can't stand behind it, I call it bullshit numbers." Alicia (exec) wants ROI; Nitin does not, and he is the customer. Accepted value framing is workflow before → workflow after with one compressed cycle-time metric.
• The real gap is FORWARD visibility (roadmap/priorities), not retrospective reporting. Trigger incident: Brandon asked Nitin whether legal work was blocked until September; Nitin couldn't answer.
• Nitin's literal ask: epic-level visibility, "point me to something... always up to date here." Plus run it as a 60–90 day experiment.
• Two distinct audiences: E-team (short, specific, launched-X-changed-Y) and org-wide (awareness so people engage rather than build shadow AI).
• Samuel Adams is being promoted to manager of AI Ops (pending finance/approvals as of the 1:1).

Solution decided: Jira APMT stays source of truth with a reportable label + structured epic template (business owner, beneficiary team, problem, before-state, target outcome, delivery stage, expected launch, outcome-at-close). Supabase Edge Functions ingest APMT + AIPM; GitLab CI scheduled pipelines are the trigger/audit log; Next.js on Vercel + Okta is the roadmap front end (standalone first, fold into AI Hub later). Monthly exec memo (narrative, 3 items, before/after appendix) to #ai-steering-committee C0A8DBG2ZD0 + E-team email; weekly internal digest incl. AIPM service-desk load. Approval loop in Slack; approvers Samuel Adams, Sam Irizarry, Marco Bettineschi.

Spec at: ai-ops-leadership-reports/SOLUTION_SPEC.md