01What actually gets scored
We score how an idea reads in one line, with no context. The model sees the one or two sentences you submitted and nothing else — no team, no traction, no funding, no market research, no idea of whether it already exists as a real, working business.
That cuts both ways. A working business can score low if the one-line description doesn’t make its case. A weak idea phrased sharply can still score honestly low, because the rubric below scores the idea, not the writing. Nothing here is a verdict on you, your execution, or a company you may have already built — only on what the sentence itself demonstrates.
02The five dimensions
100 points, split across five questions. Each is scored independently and added together for the total.
Originality25 points
How many times this has already been built. An idea that exists in several funded companies scores near zero here, however well it is executed.
Willingness to pay25 points
Whether a stranger would actually hand over money for it. What people say they want and what they pay for are different things.
Weekend copy risk20 points
Whether a competent developer could rebuild the core of this in two days. Scoring high needs a real moat: proprietary data, network effects, regulation, or genuine technical difficulty.
Real problem20 points
Whether this hurts someone today, often and expensively. It also has to still be unsolved. If an established service already handles it well, this scores low even when the pain is real.
Delusion index10 points
How much of this depends on things that will not happen. Ten means every assumption is reasonable. Zero means it needs people to change ingrained habits, or a large incumbent to cooperate.
03The same idea always returns the same score
Submit the exact same idea text twice and you get the exact same score, verdict and breakdown both times — it’s looked up, not re-guessed. Every score is cached against the normalized text of the submission the first time it’s scored, and every later identical submission returns that same stored result, permanently.
Change the wording — even lightly — and it’s a different submission to the scorer, so it can land a point or two differently even when the underlying idea is the same. The guarantee is about the text you actually submitted, not every possible way of phrasing it.
04What the score is not
It is not business advice, investment advice, or a valuation. It is one automated opinion, produced against a published rubric, about how a single sentence reads. Treat it as a blunt first reaction, not a verdict on whether to start, keep running, or shut down anything.