Define the facts that matter, paste an answer from ChatGPT, Gemini, Perplexity or another AI surface, and check for missing or conflicting brand claims.
AI Said What? is a free AI accuracy audit tool for checking how generative AI systems describe a brand. Instead of only tracking whether a brand appears, it compares the answer with a small truth set you define.
An AI brand accuracy audit records what an AI system says about a business and compares those claims with known brand facts. Useful truth-set items include products, services, locations, pricing rules, supported markets, eligibility and other facts customers could act on.
A brand can be visible in AI-generated answers and still be described inaccurately. Use Signal Sample for sampled presence and citation tracking; use AI Said What? when the problem is whether the answer reflects the important facts.
Each truth rule contains expected terms and an optional known-wrong phrase. The tool marks the rule present when one expected term appears, missing when none appear, and conflicting when the known-wrong phrase is found.
Keep a stable truth set, record the prompt, surface and date, and rerun important customer questions so accuracy issues can be reproduced and discussed with evidence.
It compares what AI systems say about a brand with a defined set of known facts and records missing or conflicting claims.
No. Paste the AI answer into the tool.
No. It checks against the rules you provide and still needs human review.
Stable facts that matter if AI gets them wrong: products, locations, pricing rules, services, eligibility and markets.
No backend is required for the audit.