Every idea-to-MVP timeline I read either takes six months or promises a weekend. Neither matches how founders actually ship. Six months is a funding round spent on something nobody has used; a weekend gets you a demo that falls over the moment a real person touches it. Six weeks is the honest number for a first version that a stranger can sign up to, use, and pay for.
This is the path I run with founders on fractional CPO engagements and teach inside the AI Startup MVP course. It assumes one committed founder, a working knowledge of AI build tools, and a willingness to cut scope every single week. It does not assume an engineering team.
Week 1: Problem validation, not idea validation
The first week is conversations, and the target is ten to fifteen of them with people who have the problem right now. Not friends, not other founders, not people who say the idea sounds interesting. You are looking for a specific, repeated, expensive frustration and evidence of what people currently do instead.
The three questions that carry the week: when did you last hit this problem, what did you do about it, and what did that cost you in money or hours? If the answer to the first is 'not recently', you have found an idea with no pull. That is a good outcome in week one and a catastrophic one in week five.
The deliverable is a one-page brief: user, problem, current alternative, the wedge, and the one metric that would prove the wedge works. If you cannot write that page without hedging, you are not ready to build, and no amount of shipping speed rescues it. The free product-market fit score is a structured, ten-question version of the same exercise if you want a scaffold.
Week 2: Product definition and the kill list
Week two turns the brief into a build plan. Sketch the user journey end to end, then split every feature into must-have for the first paying user, and everything else. The everything-else column is the kill list and it should be at least three times longer than the must-have column. If it is not, you have not been honest yet.
This is also where the AI stack gets chosen, and those choices are product decisions, not infrastructure ones. Model quality, latency, cost per action and context limits all show up directly in the user experience. A response that takes eleven seconds is a different product from one that takes two, regardless of how good the output is.
Write the evaluation plan before writing code. For an AI feature, that means a set of twenty to fifty real inputs with a defined view of what a good output looks like. Without that set, week four becomes 'it feels better now' and you will ship a regression without noticing.
Weeks 3 and 4: Build, with AI as the co-founder
Two weeks of building, deploying to a real URL every day. Lovable, Cursor and Bolt all get you from empty repo to authenticated app with a database faster than a contract team can send you a statement of work. The constraint has stopped being how quickly you can produce screens; it is whether the screens are the right ones.
Rules that keep these two weeks honest. Deploy daily, even when it is ugly. Do founder-led QA every evening on a phone, not just a laptop. Instrument analytics on day one so week six has data instead of opinions. And put the payment or waitlist step in during week three, not week six, because the flow around money is always harder than it looks.
Expect to cut something in week four. Every build does. The kill list from week two is what makes that a five-minute decision instead of a two-day argument.
Week 5: The launch surface
A working product with no launch surface converts nobody. Week five builds the rest of it: a landing page that states the problem in the customer's words, pricing you are willing to defend, a signup or checkout path that takes under ninety seconds, an onboarding email sequence, and tracking that ties signups back to source.
Pricing is the part founders defer and should not. Charging from day one changes who signs up and what they tell you. Free users give you polite feedback; paying users tell you what is broken. If you genuinely cannot charge yet, take card details for a trial rather than shipping a permanently free product.
Prep the channels this week too. One warm channel you already have, one public post, and a short list of people from week one's interviews who asked to see it when it was ready. Those interview contacts are the highest-converting launch list you will ever have.
Week 6: Launch and the first honest read
Launch publicly, get the first paying users, and then resist the urge to immediately build the next thing. The value of week six is the read it gives you: how many people who landed signed up, how many who signed up completed the core action, and how many came back in the following seven days.
Return usage is the number that matters. Signups measure your copy; the second session measures your product. If nobody comes back, no roadmap fixes it, and the correct next move is another round of conversations rather than another sprint.
Close the six weeks with a written retro: what the data says, what you cut and whether you were right, and the next six weeks stated as bets with metrics attached rather than features with dates attached.
What this costs, and what it does not buy you
Done by a founder with AI tooling, the direct cost of these six weeks is usually a few hundred pounds a month in subscriptions and hosting. Done with help, it sits in the range the MVP cost calculator walks through. Either way it is an order of magnitude below the agency quote for the same scope.
What six weeks does not buy you is a finished company. It buys you a real product in real hands and the first evidence about whether the thing is worth continuing. That is the entire point. The failure mode I see most often is not building too little in six weeks; it is spending six months building the thing week one would have killed.
Frequently asked questions
- Can you really build an MVP in six weeks?
- Yes, for a first version with one core journey, authentication, payment and analytics. You cannot build a platform with five user types and an admin suite in six weeks, which is why week two's kill list matters more than the build itself.
- Do I need a technical co-founder to follow this path?
- Not for the first version. AI build tools let a non-engineer founder ship a working, deployed product. You will need engineering help once you have paying users, load and integrations, but hiring that before validation is what burns the first round.
- What if week one kills the idea?
- That is the cheapest possible outcome. You have spent one week instead of six months. In practice most ideas do not die in week one, they narrow — the problem turns out to be real for a smaller, sharper group than you assumed.
- How is this different from the five-day version?
- The five-day sprint produces a working prototype for a single journey. The six-week path adds real validation up front and a launch surface at the end, so you finish with paying users rather than a demo.