A score you can disagree with — and inspect.
BuildCheck separates evidence collection, interpretation and scoring so an AI provider can never manufacture market proof or choose the final number.
1. Collect
A DataSourceProvider returns explicit market signals, competitors and channels. In V1, every record comes from a curated demo scenario and is marked isDemo=true.
2. Interpret
An AIProvider may structure the idea, classify supplied signals and draft recommendations. With no signals, it must return Not enough evidence.
3. Score
Pure TypeScript functions score seven dimensions: Demand 25, Pain 20, Willingness to pay 15, Distribution 15, Competition gap 10, Build simplicity 10 and Defensibility 5.
4. Decide
Code maps 80–100 to BUILD, 60–79 to VALIDATE FIRST, 40–59 to PIVOT and 0–39 to KILL. Confidence is calculated separately from evidence coverage and quality.
No live-source claims.
Names such as “Demo interview”, “Demo app review” and “Demo pricing study” describe scenario categories, not completed research. Quotes in the UI are product-writing examples. They must be replaced record by record as compliant providers are connected.
Run a demo analysis