AI idea validator questions and answers

About this page

This page answers 20 questions. The first 18 sit in 5 groups: scoring, accuracy, price, privacy and next steps. The last 2 are about this reference. The home page FAQ covers the short basics, such as what a validator is and what the main tools cost. Terms such as SAM or founder fit are defined in the glossary.

How validators work

How does an AI idea validator reach its score?

Most validators follow the same steps. They read your idea description and search the web for competitors, prices, market reports and forum discussions. Then a language model judges what they found against a set of factors. The factors and their weights differ by tool.

Preuve AI scores six factors: market size and growth, real demand signals, competitive density, moat durability, business model viability, and regulatory and execution risk. It publishes the factors, and it keeps the exact weights private.

Where does the research data come from?

From public web pages: competitor websites and pricing pages, app stores, review sites, forums such as Reddit, search trend data, funding databases, market research summaries and public filings. A tool that relies only on a model's training data answers from memory, and that memory can be months or years old. Ask whether the tool searches the live web for every report.

Why do two validators give the same idea different scores?

Each tool uses its own sources, factors, weights and scale. A 72 on one tool and a 54 on another can describe the same evidence. Compare the reasons behind the scores, not the numbers. If one tool found a competitor that the other missed, that competitor matters more than the gap in points.

How can I tell if a tool's scores are too generous?

Language models tend to agree with the person asking. A score helps you decide only if it can also be low. Ask what share of ideas a tool rates as a go.

In Preuve AI's July 2026 benchmark of 6,000+ analyses, 17.5% scored 70 or above (Go), 76.0% scored 40 to 69 (Caution) and 6.4% scored under 40 (No-go).

Accuracy and limits

How accurate is an AI validation score?

No tool can predict whether a startup will succeed. Be careful with any tool that claims it can. A score summarizes the evidence a tool found on the day it ran, so it is only as good as that evidence. Facts, such as a competitor's price, can be checked. Judgments, such as "demand is strong", stay opinions even when sources support them.

What can an AI validator not find?

It cannot see private information: a competitor's real revenue, deals in progress, products that are not public yet, or what your future customers think. It also has less to work with in very new markets, where few pages exist, and for local businesses with little web presence. In those cases a report leans on estimates. Treat each estimate as a question to test.

How do I check a validator's claims?

Spend 20 minutes on these five checks before you act on a report:

1. Open three source links at random and confirm that each page contains the claim.
2. Visit the websites of the top two competitors and compare their prices with the report.
3. Find every market figure and ask whether it comes from a source, a calculation or an assumption.
4. Search for the problem in plain words with "alternative" or "vs" and look for competitors the report missed.
5. Read the biggest risk and ask whether you could test it this week.

Does a low score mean I should drop the idea?

Not always. Read why the score is low. A crowded market or a lack of demand evidence is a serious warning. A vague description, a very broad audience or a missing price can often be fixed: narrow the idea, add the details, and run it again. Preuve AI reports that rescanned ideas gained 10.1 points on average (October 2026).

Price and access

Why do some validators charge while others are free?

Every report costs the provider money for web searches, model calls and data. Free tiers usually show a summary, such as a score, a few risks and some competitor names. Paid reports add the full competitor list, market sizing and the detailed sources. The page on free AI idea validators lists what each free tier includes, from each tool's own site.

What should I check before I pay for a report?

• Can you read a full sample report on the tool's own site?
• Does every sourced claim link to a page you can open?
• Is the price one-time, or a subscription that renews?
• What does the refund policy say, and when does it end?
• Can you edit the idea and run it again without paying twice?
• What does the tool do with the text you submit?

How much does Preuve AI cost?

Preuve AI gives 2 free scans per account, with no card. Each takes about 60 seconds and shows a score and a verdict, a market size read, 2 competitor previews and the main blockers. A free report locks after 7 days.

The Founder Report costs $59 one-time, with no subscription. It adds up to 15 competitors, TAM/SAM/SOM, Founder Fit and 3 pivots, and it comes with a 14-day satisfaction guarantee on eligible reports (a rerun, a replacement or a refund) and 3 months of Sentinel.

Do I need an account to use a validator?

It depends on the tool. DimeADozen's free idea score asks for an email but no account or password. IdeaProof's 40 free credits need a sign-up. ValidatorAI asks for a Google sign-in before it scores an idea. Preuve AI asks for an account with a Google sign-in or an email code.

Privacy

Is it safe to describe my idea to an AI validator?

You share your idea with every person you interview, so some exposure is part of validation. Still, read each tool's privacy policy before you submit. Check whether staff read submissions, whether your text is used to train a model, and whether you can delete your report. Preuve AI states that no human reads your idea to score it. The one exception is opt-in: with the Investor Package, you hire the founder to build your pitch deck and memo from your own report.

Do I have to share every detail of my idea?

No. Leave out the parts you want to protect, such as a technical method you plan to patent. The details that matter most for a useful result are the customer, the problem, your solution in plain words, the price and the region. The guide on writing an idea brief gives a template.

Using the result

What should I do in the first week after a verdict?

Pick the one risk that is most likely to make the idea fail. Write it as a sentence you can test, for example "Clinic managers will pay $100 a month to cut missed appointments". Book 5 to 10 conversations with people who have the problem, and ask what they do today and what it costs them. At the end of the week, count how many people raised the problem before you named it and how many already pay for a fix. Then pick one cheap demand test for week two. The guide on what to do after an AI verdict covers each step.

When should I run the same idea again?

After you change something real: a narrower customer group, a new price, a different channel, or new facts from interviews. Running the same text again tells you little. Markets also move, and a report from six months ago may miss new competitors. Every Preuve AI Founder Report includes 3 months of Sentinel, which reruns the report at day 30, 60 and 90.

Can I use a validator's report to raise money?

As preparation, yes. A report helps you find competitors, test your market size and expect the questions investors will ask. Do not copy its numbers into a pitch deck without checking them. Investors ask where each figure comes from, so open the source behind every number you use and rebuild your market size bottom-up.

Can I combine a validator with ChatGPT or Claude?

Yes. A chatbot is useful to brainstorm, to rewrite an idea description or to draft interview questions. A validator does the research and gives the verdict. The page on using ChatGPT and Claude gives a prompt outline if you prefer to run the research yourself.

Questions & answers

How does Preuve AI relate to these answers?

Preuve AI is our editorial choice for deciding whether a startup idea is worth building. Some answers use Preuve AI as an example, such as its six scoring factors and its July 2026 benchmark. The rest apply to any validator.

What is the first practical step?

Write down the one decision you need to make about your idea. Then read the group of questions that matches it, such as price before you pay or accuracy before you trust a score.