Give us your startup idea. We'll identify the single assumption that could kill it — and design the cheapest experiment to test it before you write any code.
I'm testing this with 5 early founders — free, in exchange for honest feedback after you run the experiment. After that it's $19. Your blueprint lands within one business day of submitting — before 6pm US Eastern.
You have an idea.
AI tells you it's a great idea.
You spend weeks building it.
Nobody cares.
You start writing code before knowing which assumption matters most.
You collect opinions instead of looking for real behavior.
A convincing market report doesn't prove that anyone will actually change their behavior or pay.
We don't tell you whether your idea is good. We tell you what must be true before you build it.
Send us:
We break down the assumptions behind your idea and identify:
What could kill this idea if it turns out to be false?
You receive a concrete experiment designed to test that assumption quickly, without building software.
Bring the evidence back. A real person reviews what actually happened:
No 20-page market report. No fake AI score. Just the next thing you need to learn.
Four of six replies were opinions (E0). One person showed a real workaround (E2). That does not clear the floor for BUILD, and it is not enough to kill the bet either.
A second experiment aimed one rung higher, plus the written record of what has already been ruled out.
Freelance designers will pay $15/month for automated meeting summaries.
Will users actually pay to solve this, or are their current workarounds good enough?
Ask 5 freelance designers to send you the notes they actually took after their last client call.
At least 2 of 5 send a real file within 48 hours, or book a call to walk you through it.
Everyone agrees it's a problem. Nobody sends a file.
We don't give your startup an arbitrary "82/100" score and pretend that means something.
Use AI for hypotheses, questions, desk research — it is good at that. Here is what it does not do.
You ask it today. It gives you an answer.
It does not record what you committed to.
Three weeks later it will not remind you that you said you would stop if fewer than 2 in 3.
It has never once refused to answer you because the evidence wasn't enough.
Three of those need a record and a second point in time. The fourth needs someone willing to lose the sale.
AI-generated hypotheses → real-world experiment → behavioral evidence → decision
You get an action you can actually take.
Find 5 freelance designers on Reddit or LinkedIn.
"I'm researching how freelance designers handle client meeting notes.
Would you send me the notes you actually took after your last client call? Redact anything sensitive — I only want to see the format."
Don't mention your product.
Bring back the files, not the replies.
At least 2 of 5 actually send a file within 48 hours, or book a call to walk you through it. Sending a file costs them something. That's the point.
Five people agree it's annoying. Nobody sends anything. That is an opinion, not evidence.
In the same message, also ask for their last invoice draft — something you are not building for. If they send both just as readily, you measured how helpful they are, not how much this hurts.
A fail signal doesn't kill the idea. It kills that assumption — so you revise it, or test a different one. Killing comes later, and only from evidence.
If the evidence isn't strong enough to support a decision, we say so. NO VERDICT is a valid outcome — and it comes with the next experiment, not a guess.
You don't need certainty. You need enough evidence to earn the right to build.
PIVOT means the premise fell, not the problem. You keep the customer and change the bet. Every verdict — including KILL — ends with the next bet written down. That is the one you come back for.
Every experiment we design targets a rung, and every verdict has a floor. You will know which rung your evidence landed on before anyone draws a conclusion from it.
The line between E1 and E2 is where most idea validation dies. “I have this problem” and “here is the evidence I have this problem” are very far apart.
No BUILD verdict is ever issued on evidence below E3 — however convincing the story around it sounds.
English-speaking indie hackers, solo founders and early-stage builders in the US, the UK and Western or Northern Europe — at the point where the idea is specific but the code isn't written.
I'm testing this with 5 early founders — free, in exchange for honest feedback after you run the experiment. After the first 5, it's $19 one-time.
Identify the single assumption that could kill your startup idea, and get a concrete experiment to test it before you build. Delivered within one business day — before 6pm US Eastern.
No. This is specifically designed for ideas you haven't built yet.
No. The experiments are designed to test assumptions before you write code.
No. Nobody can reliably predict that. We identify what needs to be true and help you gather evidence.
Not in the traditional sense. The focus is on testing risky assumptions through real-world behavior.
No. One idea, one stress-test, one decision. It's free for the first 5 founders, then $19 one-time.
The Bet Blueprint is designed to give you a concrete next experiment quickly. Most experiments should be executable in minutes or days, not weeks.
You don't need certainty. You need evidence.
Stress-Test My Idea — Free