How it works
You describe your idea and get a free quick score in seconds. For the full report, a short guided interview covers 10 areas of your idea, then an AI model writes a viability score, a go/no-go verdict, competitors, pricing, revenue scenarios and a 7-day plan to test demand with real customers.
Updated October 2026 · By the startupconcept.ai team
What you do
Write one or two sentences: who it's for, the problem and how you'd solve it. Choose your market: local, your country or global.
What you get
No idea yet? The startup idea generator starts from your skills, budget, weekly time and risk tolerance instead, and returns 3 ideas you can validate.
What you do
Nothing more: no signup needed.
What you get
The quick score is a short first read of your description. It's free so you can check several ideas before deciding which one deserves a full report.
What you do
Answer follow-up questions in a chat. The AI keeps asking until all 10 areas of your idea are clear, then asks you to confirm before it writes the report.
What you get
The interview needs 10 credits available, because it runs on the larger AI model. The credits are only used once the report is generated, and they're refunded automatically if generation fails.
What you do
Read the report as soon as the first results appear; the deeper sections fill in while you read.
What you get
A full report uses 10 credits. Credit packs start at $19.99 for 30 credits (3 reports), and credits never expire.
What you do
Follow the 7-day plan: interviews, a landing page with a real price, and a pre-sale ask. Come back and refine the report with what you learned.
What you get
This is the step that actually validates an idea. The report's job is to make sure you test the riskiest assumption first.
The full report is written by an AI language model from two inputs: your answers to the guided interview, and what the model learned about markets, competitors and business models during training. The interview covers the same 10 areas for every idea, so reports are consistent and comparable:
What a pass looks like in each area is set out in the 10-point idea viability framework.
Limits you should know
The analysis doesn't search the web in real time, so very recent launches, prices and market figures can be missing or out of date. Check any competitor or number you plan to rely on. Revenue scenarios are estimates with their assumptions stated, not forecasts. And no AI can tell you whether your customers will pay: that's what the 7-day plan is for.
Can
Can't
Want to do the whole process yourself? Our guide on how to validate a business idea covers interviews, smoke tests and pre-sales, and these real examples show how well-known companies did it.
An AI idea validator questions your idea the way an experienced founder would, then scores it. On startupconcept.ai you get a free quick score from a short description; for a full report, a guided interview covers 10 areas, and an AI model writes a viability score, a go/no-go verdict, competitors, pricing, revenue scenarios and a 7-day plan to test demand.
Yes. The quick score is free and needs no signup. It gives a 0–100 score, a verdict and the top 3 risks. There's a daily limit per visitor to keep it free.
The quick score takes about 10 seconds. The guided interview takes 5–10 minutes, the first results of the full report arrive in about 60 seconds and the deeper analysis finishes in under 10 minutes.
No. The report is built from your interview answers and what the AI model learned about markets, competitors and business models during training. Very recent launches, prices or market figures can be missing or out of date, so check any competitor or number you plan to rely on.
No. AI can do the desk research quickly and point out weak assumptions, but only customers can prove demand. Use the report to decide what to test, then run the 7-day plan: interviews, a landing page with a real price and a pre-sale ask.
Your idea isn't shown to other users or sold. It's sent to our AI provider, OpenAI, only to generate your analysis; OpenAI doesn't use API data to train its models by default.