Ask first what would end it, not whether it is good. WhittleOS runs every idea through a fixed list of deal-breaker questions, and a hard failure on any one drops it whatever else is true. The deal-breaker self-check below puts the same deal-breakers to you and combines your answers by the same gate rules; a clean result earns a scored check, not a yes.
Answer for your idea and your situation. Your answers stay in this tab — none of them is sent. Each line under a question is the short definition; the full check applies a few of them with finer rules.
Fails if there is no plausible path to >$19/mo recurring or >$100 one-time.
Fails if there is no plausible recurring revenue model and no high-volume one-time model.
Fails if the target buyer cannot tolerate $19+/mo.
Fails if there is no identifiable repeatable channel within the founder's reach.
Fails if the product requires onboarding calls, white-glove support, or manual data work.
Fails if the product depends on a single platform/API/marketplace with no exit. A no here is not meant to end the idea on its own.
Fails if the product collapses when a major LLM ships the feature natively.
Fails if the founder cannot reasonably sell, operate, or speak to the buyer.
Fails if revenue requires SOC2 / HIPAA / financial licenses or equivalent up-front.
Fails if the buyer requires legal review, an MSA, or a sales call before buying.
Fails if there is no plausible <12-month path to break-even on hosting + AI costs.
Fails if the idea is entirely about the founder's personal identity (cannot be sold/transferred).
0 of 12 answered. A hard “no” on any of them settles it at once; otherwise the result appears when every question has an answer.
A good idea is one nothing kills
Most advice on this question weighs pros against cons and asks how you feel about them. A gate works the other way round. Our idea check asks 12 questions, each phrased so that yes passes, and one hard no ends the idea: 11 passes do not outvote it.
Two rules sit on top. Platform dependence is the one soft check: the gate is built not to drop an idea for it alone, but a no there rules out "Build it", and the best call left is "Test it first". And 3 or more questions nobody can answer yet put the idea "On hold": too little is known to drop it or to back it.
One check moves with your resources. For a funded team working on it full time, some named compliance costs — a security audit, card-payment rules, identity checks — count as a costed risk that rules out "Build it" but not the idea. Regulated fields like medicine stay hard either way, and for anyone else all of it stays hard. The self-check assumes the hard case.
Only an idea that clears the gate gets scored: 7 weighted areas, a Whittle Score computed in code from fixed weights, and a call shaped by your profile. The self-check stops before that step, because scoring needs a model and evidence. The check also applies a few deal-breakers with finer rules than the one-line definitions shown in the self-check: hands-on support, for one, counts only the work that lands on you.
Your own answers are the weak part
The questions are not hard. Answering them about your own idea is. A founder who likes an idea answers generously, and a chatbot asked the same thing often agrees with whoever is asking. A self-check is a starting point: good at finding the question you cannot answer, weak as a verdict.
The idea check answers the same deal-breakers from your idea and your founder profile — your hours, your money, your skills, whether you will sell — and gives a reason for each, so you can argue with a reason rather than with a mood. It costs 1 credit, and a new account gets 2 free.
What a clean result does and does not mean
No deal-breaker means the idea may be scored, not that it is good. "Build it" is the call our check guards hardest: it needs a high score, strong evidence on several fronts and a confirmed way to reach buyers.
So the next step after a clean result is the cheapest test of your weakest answer, and that is usually whether anyone pays. A price on a page, a payment link or a pre-order tells you more than another opinion, including ours.
What the check has done with ideas whose endings are known
On 2026-06-23, on an earlier version of the gate, we ran the idea check on 8 anonymized companies whose endings are known, each described as it looked at the start. For a solo founder it dropped 4 of 5 failures and 1 of 3 successes; a dropped success means "not for one person", not a bad company. The cases are hand-picked with hindsight in them, re-running moves the ratings, and we claim no accuracy: this is what the check did on these cases, not proof that it is right.
Common questions
How do I know if my business idea is good?
Try to kill it first. Put it through the deal-breakers — can it make money, charge on repeat, reach buyers, run without hands-on support — and only then weigh how strong it is. A good idea is one that survives every hard check for your situation and then shows evidence that someone pays.
Will my business idea work?
Nobody can tell you before someone pays for it, and no tool can be measured on ideas that were never built, ours included. What a check can tell you is whether the idea fails something that would end it for you, and which answer is weakest, so your first test goes there.
Is the self-check on this page the same as the idea check?
It puts the same 12 deal-breakers to you and combines your answers by the same gate rules: a hard no drops the idea, a no on platform dependence alone rules out "Build it" instead, and 3 or more unknowns put it on hold. The difference is who answers. In the idea check a model answers from your idea and your profile, applies a few checks with finer rules, gives a reason for each, and then scores what survives.
Are my answers saved anywhere?
No. The self-check runs in your browser and none of your answers is sent to us. Close the tab and they are gone.
See it rather than take our word
Everything this page describes is published on this site, with no signup — a complete run, a full report, and the checks that produced them.