Steerco Action-Item Closure-Probability Grader for PMO Leads Running Cross-Functional Programs

A browser-based grading pack that PMO leads run before locking each weekly/biweekly steering-committee meeting's RAID log. Paste the proposed action items (owner, due-date, one-line context) and get back a closure-probability score per item, grounded in the program's own historical commitment-miss pattern: vague verbs ('align stakeholders', 'drive adoption'), missing definition-of-done, owner with a chronic slip-history on this program, items semantically identical to prior steerco commitments that rolled three meetings without closing. AE's adversarial multi-model debate stress-tests each commitment's language against the program's miss-pattern history; structured constraint language tracks lifecycle states (logged → in-flight → claimed-done → verified-at-next-steerco) with promotion/demotion/kill rules. PMO leads are already spreadsheet-and-RAID-log-native — they currently re-key action items from Teams/Zoom notes into Smartsheet/Asana/Excel by hand, so a paste-in browser form adds zero friction. Closure verified by PMO marking outcome at the next steerco; recurrence verified when the same blocked-objective reappears. No integration to enterprise PM tooling required.

PMO leads already do manual commitment re-keying; AE's adversarial grading + lifecycle states slot into that workflow with zero integration and an objective per-meeting feedback loop.

Still in evaluation (phase: ranked). No verdict yet.