FixLoop prevents apartment maintenance teams from repeating failed repairs by turning technician field notes into verified, human-governed AI memory. How It Works: The 3-Step Workflow Investigate: A technician logs a recurring issue (like a ceiling stain), gets an AI-assisted inspection recommendation, and records what actually fixed it. FixLoop drafts a reusable lesson tied directly to that work order. Verify: The AI cannot learn on its own. Every drafted lesson sits in a pending state until a human supervisor reviews the evidence and approves where it applies (building scope), who can access it (roles), and when it expires. Prove: When a similar complaint comes in later, even if worded differently, FixLoop runs a side-by-side comparison of the AI with Memory OFF versus Memory ON. Using MongoDB Atlas Vector Search, the memory-enabled AI retrieves only approved, unexpired lessons for that building, cites the exact work order, and shows judges honestly whether that memory changed the recommended action. Built-In Governance & Trust The Memory Trust Panel: Gives supervisors full visibility into every lesson's source, approval status, scope, and expiration date. Instant Revocation & Forgetting: Supervisors can revoke a lesson or wipe its AI embeddings at any time while keeping the original work orders and audit logs completely intact. Strict Backend Guardrails: Code-level policy checks verify permissions and expiration dates before and after every search, ensuring revoked or outdated data never leaks into a new recommendation.
Team
