Eight weeks. One sprint. A real answer - not another advice.

  The most important work happens before the sprint clock begins. We agree on a written proof definition — specifically, what outcome from which decision-maker, using what budget, counts as proof. Without this, the end of the sprint becomes a negotiation. With it, it becomes a fact.

We also confirm the sprint team: typically one or two people with domain expertise and decision speed. Not a committee. Not a task force. A small team with real authority and real accountability.

Two days of structured problem work — not brainstorming. We use a sequence of tools to excavate the problem from multiple angles: who feels it most acutely, who has budget to solve it, what the upstream and downstream consequences look like, and what a solution needs to do to be commercially viable.

The output is a problem statement precise enough to build from, a named buyer role, and a scope boundary that prevents the MVP from expanding beyond what two weeks can realistically produce.

  The team builds. The discipline here is PRD-first: write a product requirements document before touching any tool or model. This is the single practice that separates MVPs that teach you something from MVPs that consume two weeks and answer nothing.

We provide the AI tool stack, the development methodology, and weekly checkpoints. The team provides the domain knowledge and the decision-making authority.

  The MVP goes in front of real users or buyers from week three. The proof phase is not about refining the product. It is about testing the commercial assumption: will the right person commit budget from their own P&L to develop this further?

If the answer is yes, the sprint has produced proof. If the answer is no, we document why, extract the learning, and kill the venture cleanly. The sprint playbook and assumption log become organisational assets either way.