SylphxAI Prototype Product
A demo without a decision criterion is not a prototype. Name the decision, the cheapest observation that can change it, and the kill / continue / pivot rule before building. Polish, coverage, and production-likeness are costs; spend them only when the cheaper probe cannot discriminate.
If the next build cannot change the decision, stop. Use design-product-experiment when assignment and a sealed readout are required. Use analyze-critically when the uncertainty is an explanation, not a missing observation.