A six-week guided sprint for first-time founders and side-project builders who keep stalling before launch. Scope, build and publish one small product, with a checkpoint every week. Free, and no account needed.
0 of 24 tasks done
Week 1
Choose the small thing
One sentence and one cut list, chosen today.
Checkpoint One sentence and one cut list, on paper, today.
Give week 1 to your AI agent
You are a product coach working with a first-time founder. My domain is [domain]. My rough idea is [one line]. Help me scope the smallest version a stranger would actually use in the next two weeks. Ask at most three questions, then give me: the one-sentence pitch, the core loop in five steps, and the cut list (what this version deliberately will not do). Nothing beyond that.
Week 2
Map it before you build
A build brief your agent can execute, and a date it goes online.
Checkpoint A build brief and a date, written down.
Give week 2 to your AI agent
You are a product engineer planning a build with me. Product: [pitch]. Audience: [audience]. Give me a build brief I can act on today: the core loop in five steps, the smallest data model it needs, the stack you would pick and why in one line, an explicit cut list of what this version will not do, and a realistic date this can be online. End with the order you would build it in and the single riskiest assumption.
Week 3
Build the ugly one
A live URL with the loop working, seen by nobody yet.
Checkpoint A live URL where the whole loop works, ugly or not.
Give week 3 to your AI agent
You are a product engineer. Build the smallest end-to-end version of this product: [pitch]. Stack: the simplest one that deploys to a single host. Requirements: the core loop works end to end, no placeholder pages, no dead buttons, readable on a phone. Ignore visual polish. When you are done, list what you deliberately deferred and the one thing most likely to break first.
Week 4
Make it make sense
A stranger can finish the loop without you in the room.
Checkpoint Someone who is not you completed the loop unaided.
Give week 4 to your AI agent
You are a product engineer doing a first-run pass on [product]. Read the current code and copy, then: (1) make the empty state tell a brand new user exactly what to do first, (2) remove one feature nobody asked for, (3) fix the first place a stranger would get stuck. Show me the diff, not a plan.
Week 5
Show it to ten people
Ten users who are not you, and three real conversations.
Checkpoint Ten people who are not you have used the thing.
Give week 5 to your AI agent
You are a launch planner. My product: [pitch]. My person: [audience]. First, the metadata: the title tag and meta description a stranger would actually search for, plus Open Graph and Twitter card tags. Then list ten specific places my person already spends time, with an honest angle for each, and draft the top three posts in my voice: useful first, no hype, no fake urgency.
Week 6
Publish the story
The launch is public, and you already know what is next.
Checkpoint The launch is public, and the next sprint has a name.
Give week 6 to your AI agent
You are an editor. Help me write the build story for [product]: what I cut, what broke, what I would do differently. Keep my voice, keep it under 800 words, no hype. Then draft the post for [channel] and a short reply for the three most interested people from the launch week.
Ground rules
One product. One thing a stranger would use, built in a fortnight, spread over six weeks.
Ugly and live beats pretty and unfinished. You can make it nice once somebody has used it.
Every week ends in a checkpoint. If the checkpoint is not met, the week is not done, and the next week does not start.
Publish even when you are not proud. The point is the finished release, not the applause.
No crypto, no gambling, no pay-to-win. Not even as a joke.