How it works
Four steps. Nothing hidden: the research, the questions, and every response stay visible to you the whole way through.
01
Describe what you're building
A paragraph is enough. What the situation is, and what goes wrong today. Not the whole product if only one feature is uncertain: describe just that part.
- Attach a deck, notes, or a link and we read them for context
- Tell us who it is for and where they are, so research looks in the right market
- Your idea is saved before any agent runs, so nothing is lost if a step fails
02
We research whether the problem is real
Real search, not a model guessing from memory. We name the products that already solve this and the gap they leave, and we surface the case against the idea as deliberately as the case for it.
- Every claim links to where we found it
- What people do instead today, which is usually the real competition
- The strongest counter-evidence, kept in its own section so it cannot be buried
- When live search turns up nothing, the report says so rather than inventing sources
03
You get answers from real people
Two routes, one report. Gather answers yourself with questions we write and a link you share, or hand the whole thing over and have interviews sourced and run for you.
- Questions built from your research, editable, in your own words if you prefer
- A public link that needs no signup to answer
- Every response screened for quality before it counts toward anything
- Both routes feed the same pool and the same score
04
You get a score, and the reasoning behind it
Never a bare number. The confirmation rate, the themes across every response, the risk factors, and the raw responses themselves are all there to check yourself. If the sample is too small to trust yet, the report says so plainly rather than rounding up.
Invoice chasing for freelance designers
Go ahead14 of 19 people described chasing invoices unprompted, and six already pay for something to help.
The strongest objection: most said they would not switch tools mid-project, which makes onboarding timing the real risk.
- Confirmed
- 74%
- Responses
- 19
- Sources
- 11
- Round
- v2
Every claim links to where we found it
Sample skews to one community, flagged as a risk factor
All 19 raw responses stay readable, not just the summary
An illustrative report. The numbers are made up, not a customer’s.
The final call is always yours. Proceed, rework, or stop: the product records the decision and the reasoning behind the number, but it does not make the decision for you.
Validation is a loop, not a verdict. If the signal is weak, sharpen the idea and run it again. There is no limit on rounds, and every version you have been through stays readable.
A signal, not a guarantee
A high score means the evidence so far points to a real problem people will talk about and pay to solve. It cannot promise the product will succeed, and a low score is not a verdict on you.