Run your product the way a good founder would — from the bet to the next evidence-backed decision.

Question one — What are we really betting on?

The first four milestones turn a raw idea into an explicit bet: premise, customer, problem, alternatives, and wedge — before committing real effort. This is where most AI tools skip straight to generating code; Motriz starts where a founder would.

Milestone 0
Where are we starting?

Get the lay of the land: what already exists, what needs attention now, and what constraints shape the work. Existing research, an existing repo, a product already in the market, or nothing at all — the journey starts from your real situation.

Milestone 1
What is the bet?

State the premise clearly enough to challenge it. The product thesis is explicit and falsifiable, and it gets compared with the honest alternatives — a manual process, narrower software, a more ambitious play, or doing nothing at all.

Milestone 2
Who feels this, and how?

Name a reachable customer with observable pain — not a persona-shaped guess. Problem-framing and customer-profile specialists look for the pain or the workaround people already live with.

Milestone 3
Why this opening?

See what people do today and where a credible wedge is — one proportionate to what you want this to become. A market memo and competitor map keep you honest about why anyone would switch.

Question two — What must be true?

Milestone 4
What would change our mind?

Rank the riskiest assumptions and design the cheapest tests that could change the plan — with explicit stopping rules set before the results come in. That's the validation roadmap: what you'll test, what result means what, and when you stop.

Question three — What did the market do?

Milestone 5
What did people actually do?

Run the tests and face what people actually did — not what they said, and not what you hoped. Every result lands on the journey as evidence, and each linked assumption is updated: supported, contradicted, or inconclusive. No grading your own homework.

Question four — What is the next best bet?

Choose the smallest useful move that attacks the riskiest remaining uncertainty. Sometimes that's code. Sometimes it isn't.

Milestone 6
How could this win?

Turn the evidence into a sharp plan: the wedge, the buyer, a reachable channel, a pricing hypothesis, and the risks still open — a strategy memo, not a vibe.

Milestone 7
What is the cheapest useful build?

Not every product needs an MVP first. A concierge test, a prototype, a technical spike, an MVP slice — the build is sized to the learning goal, and it ships with a learning plan: who you're putting it in front of, what the value event is, and what result determines the next move.

Milestone 8
Continue, build, strengthen, reshape, or stop?

The build checkpoint. The co-founder prepares a scorecard — evidence for, evidence against, open risks — and recommends one of five recorded decisions: Continue, Proceed to build, Strengthen the evidence, Reshape, or Stop on evidence. You make the call, and every verdict is recorded with its reasoning — override the recommendation and both positions stay on the record. If the answer is stop, you've recovered the months a doomed build would have burned — and everything you proved is yours to keep.

Question five — Is real pull forming?

The journey doesn't end when the code ships. After launch, what the market actually does goes on the record — on the same line, with the same standard of evidence. Record signals manually or connect Amplitude or PostHog, and your co-founder is ready to hear how each ship landed.

Milestone 9
Are people reaching value?

Launch the build and watch whether the intended cohort actually reaches the value event — activation as an observed fact, not a launch-day feeling.

Milestone 10
Do they come back or pay?

Look for repeated value, willingness to pay, the reasons people churn, and what delivering it really costs you.

Milestone 11
Can pull repeat?

Test whether acquisition and delivery can repeat without hiding the constraints. One repeatable motion beats three lucky spikes.

After the build call: the handoff

When the verdict is Proceed to build, the product moves into an active build — same product, same journey, same co-founder. A build pack travels into your repository: the learning plan, only the preparation the chosen build type needs (an MVP slice brings the MVP scope, a lean PRD, and UX states), and an evidence pack digesting the interviews, tests, and checkpoint verdicts that justified the build. Raw documents never enter the repo; the co-founder keeps the full memory.

The scoped build seeds the first tasks, queued riskiest-first. Coding agents — Claude Code, Codex, ChatGPT and other leading providers, running on the subscriptions and keys you already have — execute them one at a time or in parallel isolated Git worktrees. Every finished piece of work walks you through intended outcome → changes → checks → risks — and ends with your call, not the agent's. Nothing runs until you press Run, and every task opens already knowing the product: the evidence, the approved direction, the spec. The co-founder comes with it — and stays for the part after launch.