Community
Board Excellence Companion
A readiness, evidence, and scoring companion for a national nonprofit's advisory boards, where the AI refuses to claim a standard is met without a record behind it.
01 The story
National board-certification programs run on binders and memory. This companion gives a board a live dashboard of exactly where it stands: a dual-mode scoring engine computes Silver and Gold readiness, members self-attest attendance and giving, a chair-level queue verifies each claim, and evidence uploads live in a private bucket with expiring signed links. The grounded Q&A is the honest part: the assistant answers from computed status and the knowledge base, and returns unknown rather than asserting an unmet standard.
02 What went into it
- Dual-mode readiness scoring with a computed drop standard
- Member attestation with a verification queue
- Evidence uploads in a private RLS-gated bucket
- Grounded Q&A that never asserts without a backing record
- Audit viewer and review-gated export packages
- Role-scoped access: member, chair, governance, officer, admin
03 Claude in production
Every AI call runs server-side under the Lumen governance layer: no client keys, cost telemetry on every call, and a human confirming every consequential write.
04 Stack and capabilities