Beyond the Demo
What the demo shows: it works. What it hides: how well it scales, what breaks under load, how much it costs per user, how dependent it is on specific people.
Key assessment areas: Architecture scalability (can this handle 100x users?), Team bus factor (who can't leave?), AI unit economics (is the AI feature profitable?), Dependency risk (single vendor lock-in?), Testing maturity (what happens when something breaks?).