Before you waste 3 months building — know what to kill, validate, and build.
A decision tool for one-person businesses. It searches, it drops what it cannot tie to a problem it found, and it is built to say no. Start wherever you actually are.
I have an idea
Get a straight answer: build it, test it, or drop it — with the reasons attached.
- Describe it. Four short questions — a name, the idea in one line, who it is for, and the pain. No document, no deck.
- 12 checks that can kill it. Any one of them ends it. What survives is scored on 7 areas, weighted for what one person can build, sell and support.
- You get a verdict. Build it, test it, or drop it — with the reasons, the biggest unknown and one next move. Under a minute, and it is willing to say no.
Free with the 2 credits you get on signup — no card required.
I don't have one yet
We'll go and find you some — from real market data, not a model's memory.
- Tell us about you. A minute, free, no credits — skills, time, budget, reach. The search is ranked against these answers, which is why it comes first.
- We search a market. Real sources, and anything with no proof that people have the problem is dropped. Across 30 markets that meant keeping 93 of 3,430.
- You get up to 5 ideas. Each with the evidence behind it, a build / test / drop call, and one first move. A thin market honestly comes back with none.
Step 1 is free. The search itself is from 2 credits, which the signup credits don't cover — see a real run first.
Either way, this is what comes back
A verdict, a score for one person, the reasons behind it, and one thing to do next — plus the point where you should walk away. Door one returns this for the idea you bring. Door two returns up to 5 of them, each with proof that people really have the problem.
The card beside this is real output from the run we publish, unedited — including the score, which is a 52 and not a 90.
SoloPipe — the best idea the run we publish came back with, scored for one person.
Next move
Ship a narrow landing page targeting the single-user pipeline wedge and test organic community response before writing code.
Real output, unedited. Read the whole Discovery run.
You can ask a chatbot for criticism. You can't ask it for a decision.
- A chatbot
- Here
- Answers from memory
- Runs live searches and reads the pages
- Invents competitors that sound real
- Drops anything with no proof of the problem
- Nothing you can open
- 32 of 56 problems carry a link
- Grades you differently every time
- Same fixed checklist every run
- Starts from zero each session
- Each run builds on what earlier runs verified
As of 2026-06-10, a leading validator scored our own idea 72/100 — “PROMISING 🚀.”
We ran the same one-liner through our own tools: Kill My Idea → KILL, 15/100, with 10 specific reasons. Single Idea Check → KILL, 0/100 (KF-07 failed: “remove the LLM, what's left?”).
Dated and screenshotted — a real run, on a real day.
The rest are marked not verified, and the page says which. Browse the source-linked pain library, the live feed of problems, and the scored idea library.
What actually runs when you press Run.
If you bring an idea
seconds- Your idea
- 12 checks that can kill it
- Scored on 7 areas
- A verdict
- Any one of the checks ends it. Nothing is searched — this path judges what you brought.
- The model rates each area. The total and the pass/fail line are computed in code, so it cannot hand itself a passing number.
- The verdict is one of Build it, Test it first, Try a paid pilot, Not yet or Drop it— never a number on its own, and never a maybe.
If you want ideas found
10–40 min- Your market
- Split into niches
- Live search
- No proof of the problem? Dropped
- Checks that can kill it
- Scoring
- Up to 5 ideas
- Three kinds of source: a search API, our own page reader (which respects each site's robots rules and never follows a link the model invented), and public APIs. Every run lists the sources it actually read.
- Runs in the background — we email you when it's done. The long version.
One search, start to finish
That is the run we publish, end to end: 20 niches planned, 60 searches, 24 pages read in full, 41 sources you can open — and each of the 36 it dropped says which check killed it. One market on its own is capped at 150 ideas.
Don't take our word for it
Every number above comes from one published run, and the whole thing is open — the sources it read, the ideas it dropped, and why. No account, no payment.
See a real Discovery runLonger reads
Where ideas actually come from, what a check that can say no does differently, and the questions people ask before trusting any of it.
- E-commerce startup ideas from merchant pain — and which ones we would kill
- Low-cost startup ideas for solo developers, ranked by time to evidence
- Search intent vs. search volume: how we chose what not to publish
- AI startup ideas: the wrapper question is not what kills them
- Am I the right founder for this idea? What the answer costs
- Business ideas for agencies: where to look without inventing demand
- Business ideas for coaches: what software can and cannot solve
- How to price a micro SaaS, and the floor nobody names
- Your idea is too broad. What a real narrowing has to contain
- What to build first — and the case where you should not be asking yet
- Will your SaaS be support-heavy? It is decidable before you build
- Do startup idea validators work? We tested ours and published the misses
- Famous startup failures, and the shape they share
- How many customers you need, and whether you can carry them
- How to score a business idea, with the weights published
- Is my startup idea already taken? The wrong question, and the one that works
- Platform risk, counted rather than warned about
- The startup idea validation checklist, in the order that saves time
- Validating a startup idea without talking to customers
- Where to find real customer problems, and what makes one quotable
- Why AI says your startup idea is great
- Why startup ideas fail: 36 rejections, itemised
- Micro SaaS ideas: the arithmetic that decides them
- SaaS ideas: where they actually die
- Where startup ideas actually come from
- Finding startup ideas on Reddit, and what that method can and cannot tell you
- Tech startup ideas and the question that now decides them
- Why Y Combinator startup ideas do not transfer to one person
- Best startup idea validation tools (2026): what actually matters
What a real startup idea check looks like walks through the checks one at a time. Short answers to the questions that come up first live in answers, and every guide is listed in the guide index.
Stop building the wrong thing.
Check your own idea in under a minute — or give the pipeline half an hour to go find, screen and score ideas for you. Saying no early is the whole point.
Pay for the decision, not the dopamine.