I kept reading AEO advice that was a checklist of files — add llms.txt, add schema, tick the box. But when I actually fetched the sites, the things blocking citation were never the missing files. They were a 403 at the CDN that no one had looked at, and pages whose text only exists after JavaScript runs. A tool that reads robots.txt cannot see either. So AIFindUs fetches instead of asking. It checks the response a crawler really gets, and it fetches again with JavaScript off to see what survives. The part that changed most while building it was honesty. Early versions let a model write the findings, and it produced confident sentences the evidence did not support. Now the score is computed in code — the same evidence always produces the same number — and generated claims are dropped unless they can be matched to the site's own words. The limits page is deliberately long. It is free, with no account and no API key. I would genuinely like to know what it gets wrong on your site.

AIFindUs

See whether AI answer engines can actually read your site

See whether AI answer engines can actually read your site