Tools that tell you what they cannot see.
Each one checks a single thing and says plainly where its answer stops. No sign-up, no email wall, and nothing here is scored against a benchmark we invented.
A tool reads one page. It does not measure your category.
Every tool here fetches the URL you give it and asks a language model to assess what it finds. That is useful and it is fast, and it is not the same kind of thing as a measurement. A model can be wrong about a page in ways a counter cannot.
So none of these outputs is a citation share, none is comparable with anything in a Caldrin report, and none of them queries an answer engine to see who gets named. That work is the report, it runs on a different instrument, and it carries a sample size and a window.
- instrument
- A language model reading one page you supply. Not Caul.
- what was measured
- Nothing is measured. Each tool returns an assessment of a single fetched page.
- how
- The page is fetched, its HTML extracted, and the extract sent to a model with a fixed rubric stated on each tool page.
- over what window
- A single moment. Re-running tomorrow can return a different answer, because models are probabilistic.
- what this cannot tell you
- No tool here queries an answer engine, sees a competitor, reads a page behind a login or a paywall, or renders JavaScript. None of it tells you whether AI names you, which is what the report is for.
When you want the measured version
A tool checks one thing about one page. The report measures whether AI names you in your category, against competitors you nominate, with the engines stated and the sample size attached.