Opportunity brief
Prompt Down: The AI-Native PM Challenge
Write the spec. Let the agent ship.
Most product case competitions end at a slide deck. Prompt Down goes further: you design the product, write the blueprint, and then watch it get built live by an AI coding agent, with no manual coding allowed.
What is it? A multi-stage Product Management case competition combined with a live AI build-off. Teams pitch a product idea, turn it into airtight specifications, and then use Claude Code at our live stations to generate a working app from their documentation and prompts alone.
How it works
- The Pitch: Solve an industry-grade problem statement and submit a 7-10 slide product deck covering personas, value proposition, architecture, and go-to-market.
- The Blueprint: Shortlisted teams get 24-48 hours to write a PRD, TRD, UI/UX specs, and a metrics plan. These become the agent's instructions.
- The Live Build: Teams take turns at Claude Code Stations (45-60 minutes each) and guide the agent using only their documents and prompts. Token usage, build time, and debugging loops are tracked automatically.
- The Demo and Defense: Present your deployed MVP in a 3-minute live demo, then defend your decisions in a 2-minute Q&A with a panel of product leaders, engineers, and investors.
How you're judged
- Product Strategy and Deck : problem depth, customer clarity, business viability
- AI Efficiency and Execution : token efficiency, time-to-deploy, prompt precision
- Fidelity and UX : how closely the running app matches your PRD/TRD, flow quality, KPI tracking, edge-case handling