Start with the problem, not the feature list
A founder can arrive with twenty features and still not have a clear product. The useful starting point is the user, the painful workflow, and the decision the first version needs to validate. If the MVP cannot teach you something important, extra functionality only makes the experiment more expensive.
Define the smallest complete workflow
A good MVP usually needs one end-to-end journey that works properly. That may include sign-up, the core task, a result, payment or subscription, and the admin controls required to support users. It does not need every future module on day one.
Know what has to connect
List the systems that matter early: payments, CRM, email, data providers, mobile apps, authentication, existing databases, or partner APIs. Integrations often shape architecture more than visual screens do.
Decide what must survive after validation
Some prototypes are intentionally disposable. Others should become the base of the production product. If you expect to keep the code, clarify ownership, environments, database structure, testing, deployment, and handoff before the project starts.
AI can accelerate execution without replacing product judgment
AI-assisted development can shorten research, scaffolding, testing and implementation. It does not decide what users need, where complexity should live, or which shortcuts will become expensive later. The fastest useful build combines speed with clear technical judgment.
What we usually want from a founder before development
A concise product brief, the primary user, the core workflow, examples of existing alternatives, must-have integrations, expected business model, and what success looks like after the first release are usually enough to start a productive discovery process.