RFP TrustRoom

RFP TrustRoom turns a messy B2B RFP and security questionnaire response into a coordinated multi-agent workflow. A human brings customer requirements, questionnaire rows and company knowledge snippets into the room, then specialized agents for requirement decomposition, evidence retrieval, answer drafting and compliance review collaborate through Band-style handoffs. They share structured context, pass object references, challenge unsupported claims, escalate high-risk answers to a human SME and produce a traceable submission pack. The product promise is simple and verifiable: every answer carries evidence, review, approval and final-pack decision context, so risky commitments stay out of the customer package. The demo focuses on enterprise control rather than unchecked automation. Q-002 shows scoped SME approval, Q-004 shows a reviewer challenge followed by bounded legal-approved wording, and Q-006 shows a fail-closed blocker when incident-response evidence is stale/conflicting and no valid approval exists. Public replay is the demo-safe evidence path: judges can inspect the full workflow without live credentials or raw room identifiers. Live Band mode is separately gated; Band REST room and handoff evidence were verified separately, and connected-peer autonomous replies remain pending until challenge-token smoke evidence passes. This is a hackathon working prototype, not a production deployment, formal audit log, compliance certification, legal/security advice system, or complete autonomous live Band workflow.