Recursion Lab

**FloorTruth - Listen. Verify. Improve.** Frontline workers see operational friction before it appears in dashboards: incomplete handoffs, recurring shortages, and barriers to training. Yet organizations often lose the context behind these observations before anyone can investigate the causes or measure an improvement. FloorTruth transforms frontline observations into source linked evidence cases that humans can investigate, challenge, and act on. The deployed prototype demonstrates synthetic observation capture, deterministic topic grouping, source-linked case review, live Meta Model API inference, hypothesis exploration, simulated human approval, an outcome measurement worksheet, and governance accounting. Unlike a conventional feedback chatbot, FloorTruth maintains the connection between evidence, competing explanations, human decisions, and measured learning. AI supports investigation without replacing human authority. Missing evidence, disconfirming tests, inference usage, estimated costs, and budget decisions remain inspectable. The working demonstration includes a simulated end to end workflow and a downloadable learning report containing evidence and event metadata. The prototype uses synthetic data and simulated approvals; real world effectiveness and production readiness remain future validation objectives. **Our mission:** Connect frontline knowledge with AI assisted investigation, accountable human decisions, and measurable organizational improvement. *Listen. Verify. Improve.* **FLOORTRUTH_DEMO_PASSWORD: Zork17845678321@@