Deliverable 4f · live · Claude-powered
Use-Case Intake
Describe an AI use case and the AI Governance Board convenes on it in real time. Four seats review it through their own lens, the chair synthesizes a calibrated disposition against GAO-21-519SP, NIST, and the OMB rights-impacting standard, and Independent Assurance red-teams the result.
This demo runs on Claude (Anthropic Opus 4.8). In production inside the OIG environment, every model call routes through your sanctioned gateways, AskSage (including its API) and Microsoft AI Foundry, with no external model key introduced.
Intake form
Describe a use case and the board will convene: four lenses, a synthesized disposition, and an independent red-team, all in real time.
Decision log
6 entries| ID | Use case | Tier | TLP | Disposition | Date |
|---|---|---|---|---|---|
| UC-001 | Contractor Fraud Risk Scorer | High-Impact | TLP:RED | Conditionally Approved | 2026-03-10 |
| UC-002 | Healthcare Provider Fraud Detector | High-Impact | TLP:RED | Approved | 2026-01-18 |
| UC-003 | Hotline Complaint Auto-Triage | Limited | TLP:AMBER | Approved | 2026-02-28 |
| UC-004 | Mail-Theft Image Classifier | High-Impact | TLP:RED | In Review | 2026-04-14 |
| UC-005 | Audit Report Summarization Assistant | Limited | TLP:AMBER | Approved | 2026-03-05 |
| UC-006 | Congressional Request Intake Classifier | Limited | TLP:GREEN | Approved | 2026-01-30 |