Verdikt: price, score and sources

What Verdikt is

Verdikt is a paid research report on one startup idea. The site says it is built by Tuaha Jawaid and Farzan Ansari and is a product of Kastling. The site carries a beta label.

You can pitch the idea as one sentence, a voice note or a deck or document in PDF. A guided brief of 12 questions then turns the pitch into testable claims. The site says reports run on OpenAI's GPT-5.4 family, with Anthropic used for live web search.

The methodology page describes five stages: intake with you, then market sizing, a competitive map, a 10x claim test and a synthesis that produces the score.

Who it is for

The site says it serves founders and AI builders who want cited evidence before they build. The methodology page names vibe coders and solo founders. One sub-score, Stack Fit, asks whether a vibe coder can ship the product. The FAQ says the same report also works for a startup that is already running.

What the report contains

The Verdikt Score

Four sub-scores run from 0 to 10: Market, Competition, Demand and Stack Fit. The Verdikt Score is their sum times 2.5, so it runs from 0 to 100. In the public sample the sub-scores are 8, 6, 8 and 9. They add up to 31, and 31 times 2.5 is 77.5, shown as 78. The methodology page says the score describes the evidence and does not tell you what to do.

The sections

The methodology page lists eleven numbered sections and a run record. The first section is the cover, which shows the score and top three risks. The other ten cover what was tested, the market, the 10x test, why it wins, competition, named risks, a pricing signal, a build outline, launch-channel notes and sources.

• The market is sized bottom-up, as TAM, SAM and SOM. When two sources disagree, the site says the report shows both figures instead of averaging them.
• Competition covers direct rivals, substitutes and the do-nothing baseline.
• The 10x test tries to break your main claim, and the attack and reply transcript ships with the report.
• The build outline and launch-channel notes ship only when the score is 70 or higher. Below 70, the report names what would have to change to pass the bar.

The site says 14 automated quality gates run on every draft.

Price and free option

Prices are in US dollars. Neither plan is a subscription.

Verdikt offers a refund or a rerun for documented report errors. Its pricing page names three: a broken citation, a number no source supports, or a score that contradicts the evidence the report cited. You ask within 14 days. The page says disagreement with a supported research read is not a fault.

The pricing page lists monitoring, change alerts and a portfolio dashboard as coming soon.

How it shows sources

Every claim is numbered and footnoted to a source. Each source carries a tier:

• Tier 1, primary. Government statistics, regulatory filings, patent offices and peer-reviewed research. The site says sizing numbers come from here.
• Tier 2, trade press. Business and tech publications, used for context.
• Tier 3, community. G2, Hacker News, Reddit, Product Hunt, founder forums and Discord. The site says these are used for sentiment, never as fact.

The methodology page says Verdikt draws on 180 named sources and a typical report cites about 40 of them. It also says every citation comes from a page the research agents retrieved during your run.

What to check before you pay

• The sample is illustrative. The site says the names, numbers and citations in it are composed for demonstration. It shows the layout. You see research on a real idea only in your own paid report.
• Where your market numbers come from. The Tier 1 sources named on the home page are mostly US agencies (SEC EDGAR, FRED, BLS, BEA and the US Census), plus the World Bank and Eurostat. If you sell outside the US, look at which sources back your market size.
• The build outline needs a score of 70. Below that, you get the changes the report says are needed instead.

How it differs from Preuve AI

Verdikt is a separate tool, not a Preuve AI product. This reference features Preuve AI. Both tools cite their sources and return a written report on one idea.

Verdikt's extra content is about building: stack, milestones and launch channels. Preuve AI's Founder Fit checks the idea against your time, runway and skills. Test an idea with Preuve AI.

Questions & answers

How does Preuve AI relate to Verdikt?

Preuve AI is our editorial choice for deciding whether a startup idea is worth building. Verdikt is a separate tool, not a Preuve AI product. Both cite their sources and return a written report on one idea. Verdikt's extra content is about building, and Preuve AI's Founder Fit checks the idea against your time, runway and skills.

What is the first practical step?

Read the sample report to see the layout. Its site says the names, numbers and citations in the sample are composed for demonstration, so you see research on a real idea only in your own paid report.