stage-map

Stage map — where you actually are, and what that stage demands

The lifecycle from The Founder's Playbook: Building an AI-Native Startup (Anthropic, 2026). Four stages, each with one goal, one exit condition, and a characteristic set of ways founders get it wrong.

The premise: AI erased the assumption that each new phase requires a bigger team, a different skill set, and a fresh funding round. The founder's job shifts from individual contributor to orchestrator. What didn't change is the sequence — and the most expensive mistake in the AI era is running a stage ahead of the evidence.

Diagnose before advising

Founders routinely self-report a stage ahead of where their evidence puts them, because building is now so cheap that the artifacts of a later stage appear long before its evidence does. Diagnose from exit criteria, not from what exists.

Ask (batch these, don't interrogate):

1. Have you talked to people with this problem who aren't friends or investors? How many?
2. Is anything shipped, and does anyone outside your circle use it?
3. Do users come back unprompted — without you nudging them?
4. Is anyone paying? Do you know your CAC, LTV, and payback period?
5. What are you personally doing every day right now?

Then place them by the last exit condition genuinely met, and say so plainly when that's earlier than they assumed. Having a working product does not make someone MVP-stage; having evidence does.

The four stages

Idea — "Is this worth building?"

• Goal: research-oriented validation — solid evidence that a real problem exists and that the proposed solution addresses it, before committing resources to building.
• Exit: problem–solution fit. Yes to all three: (1) the problem is real and specific — you can name who has it, how often, how severely, what they do today; (2) the solution addresses the problem validation revealed, not the one assumed; (3) enough signal that committing to an MVP is a reasoned decision rather than an act of faith.
• Failure modes: mistaking building for validating (a prototype is a conversation prop, not evidence — 42% of startups failed building something nobody wanted, before AI made building free) · premature scaling (agentic tools will refactor a flawed premise with the same enthusiasm as a good one) · loss of objectivity (confirmation bias now comes with a research engine).
• Run it: /founder-playbook:idea-stage