Frequently asked questions

PivotProof is an AI tool that pressure-tests startup ideas by simulating five hostile expert personas — a venture capitalist, a skeptical customer, a competitor founder, a domain expert, and a devil's advocate. Each persona critiques the submitted idea, and the system returns a single Pivot Score from 0 to 100 along with red flags, kill criteria, and a traction validation checklist. The whole report completes in roughly 60 seconds.

You submit a startup idea (title, description, target customer, and stage). PivotProof routes the idea through five carefully-prompted AI personas in parallel, each generating a verdict (REJECT / REVISIT / BUY) and a confidence score. A synthesis pass then aggregates the five verdicts into a final Pivot Score from 0 to 100, identifies the strongest red flags, suggests concrete pivots, and outputs a traction checklist of the next 3-5 validation experiments to run.

PivotProof uses one-time credit packs. The Starter Pack is $99 for 10 reports, the Pro Pack is $199 for 25 reports, and the Scale Pack is $499 for 75 reports. There are no subscriptions, and credits never expire. New users receive one free credit on signup, so you can run your first report without paying.

Your first report is free. Every new account receives one free credit on signup, which is enough to run one complete five-persona report including the full Pivot Score and synthesis. Additional reports require purchasing a credit pack (starting at $99 for 10 reports).

A Pivot Score is a single number from 0 to 100 that quantifies how survivable a startup idea looks when stress-tested against five hostile expert lenses. Scores under 35 indicate the idea has structural disqualifiers (BURNED). Scores from 35-69 indicate the idea is salvageable with a sharper wedge (ON THE FENCE). Scores 70 and above indicate the idea passes most standard hostile filters (SURVIVOR). Most ideas score in the 20-50 range on first submission.

PivotProof is built for solo founders, indie hackers, startup studios, accelerator program leads, and VC scouts who want brutally honest, structured idea validation. It is particularly useful for founders in the idea-to-MVP stage who are deciding what to build, and for operators evaluating whether to pivot or persevere on an existing product.

PivotProof uses Anthropic's Claude (currently Claude Sonnet 4.5) for all five persona critiques and the final synthesis. Each persona has a distinct system prompt engineered to override the model's default agreeableness and enforce in-character, hostile-by-default critique.

AI validators are highly reliable at catching idea patterns that historically fail — saturated categories, missing moats, weak wedges, marketplace liquidity problems, and unit-economics issues. They are unreliable at predicting willingness-to-pay, which is empirical and requires real customer interviews. Used as a pre-customer-discovery filter, AI validators dramatically improve the quality of which ideas a founder commits time to. Used in isolation as a replacement for customer interviews, they will miss the willingness-to-pay signal.

ChatGPT and most general-purpose AI chatbots are trained via RLHF to be agreeable, which makes them poor critics of startup ideas. Ask ChatGPT 'is this a good idea?' and you'll almost always get a balanced response — three positives, two concerns, an encouraging closer. PivotProof uses adversarial system prompts that explicitly forbid hedging, plus five distinct hostile personas with conflicting incentives, plus a quantitative scoring layer. The result is structured criticism that changes founder decisions rather than confirming them.

A standard report completes in approximately 60-70 seconds. The five personas run in parallel via concurrent LLM calls, and the final synthesis layer adds about 8-12 seconds. You'll see personas stream in live as the report generates.

Yes. Every report has a 'Make public' option that generates a unique share link at pivotproof.org/share/{id}. The shared page is fully public, requires no login to view, and includes an auto-generated 1200×630 OG image that previews on social media. You can unshare a report at any time.

No. PivotProof does not use submitted ideas to train AI models. Your idea text is sent over an authenticated API to Anthropic to generate the report, and per Anthropic's API terms this API data is not used for model training. PivotProof itself does not run any training pipeline on user-submitted content.

A kill criterion is a specific, dated, binary metric written in advance that, if missed, will cause you to shut down the project. Example: 'By June 1, we will have 10 paying customers at $99/month or more. If we do not, we shut down.' Kill criteria are written on day one, before sunk costs and identity attachment make objective evaluation impossible. Every PivotProof report generates suggested kill criteria specific to the idea.

Refunds are available within 14 days of purchase for unused credits, or anytime if a report fails to generate due to our error. If a report has been generated and used, refunds are not available, since the underlying AI cost has already been incurred. Email info@pivotproof.org for any billing concern.

Yes. PivotProof includes the standard five-persona panel for generic ideas, plus industry-specific persona packs for B2B SaaS, Consumer/DTC, Marketplaces, AI/Dev Tools, and Fintech. Each pack swaps the generic personas for specialists in that vertical (e.g., the Marketplace pack includes a liquidity hawk and a supply-side founder). Persona packs are $99 each, or $199 for the bundle of all five.

Yes. The Studio plan ($199 for 5 seats) and the Studio Pro plan ($499 for 15 seats) let multiple founders share credits, see each other's reports, and collaborate on validation. Team owners can invite teammates via email; invitees receive a one-click join link via Resend. Each seat has the same access to the panel and persona packs.

Because hostile is more useful at the validation stage. A real Series A investor passes on 95% of decks; their default is to find the disqualifier in 90 seconds. A real customer almost always has a workaround they prefer over a new tool. A real competitor knows exactly why your wedge is harder than it looks. Surfacing these hostile views before you spend six months building gives you the same information without the calendar cost of dozens of meetings.

No, unless you explicitly choose to make it public. By default every report is private to your account. The 'Make public' toggle adds the report to the public share gallery and generates a share link; without that toggle, no other user — including the PivotProof team — has routine access to your submission.

PivotProof's Terms of Service are governed by the laws of the United States and the State of Delaware. Privacy practices comply with general data-protection principles including GDPR and CCPA-style data-access rights. Contact info@pivotproof.org for any data request or jurisdictional question.

Yes — this is the most common and most useful use case. Roughly 70% of PivotProof submissions come from founders who haven't built anything yet, who are choosing between several ideas, or who are sharpening an idea before customer interviews. The system is specifically designed for the idea-to-MVP stage where most validation tools are weakest.