Revenue Radar

Revenue Radar turns one unstructured founder "brain dump" (pasted text — tasks, emails owed, deadlines, client asks, stated deal values) into an auditable, same-day revenue decision. A 4-stage LLM pipeline does: Extract — pull every discrete commitment with verbatim source text, dates, and stated $ values. Score — classify each into one of six revenue motions (collect / close / deliver / retain / grow / operate) and rate 1–5 on revenue proximity + urgency. Decide — partition every item into ≤3 Money Moves, Parked (safe to skip), or Blocked (waiting on someone), with a deterministic priority = revenue_proximity×3 + urgency computed in Python, not by the model. Prepare — draft only the reply/unblock messages actually needed to execute the decision. It is for solo founders / one-person companies — specifically the narrower, provable wedge where revenue evidence is concrete and provable, rather than every conceivable solopreneur. A single-page React app (no accounts, no persistence, no integrations): - One textarea + "Find My Money Moves" button, pre-seeded with a demo brain dump. - Four result sections: Today's 3 Money Moves, Blocked → Unblock, Parked Safely, All Commitments. - A Decision Receipt per item showing the motion badge, verbatim evidence, cost-of-delay, and the priority arithmetic spelled out. - A Cash Leak Radar — a visual breakdown of commitments by revenue motion. - Copyable message drafts attached to selected Money Moves / unblock requests. - Backend: FastAPI with GET /health and POST /process, deployed as one Vercel project (static frontend + Python serverless function).