Free AI visibility checker
Paste any public https URL. We fetch it the way an agent does and the way a browser does, then score the difference against the published spec. Free, no account, and every check shows its evidence.
See your site the way these assistants do
How it is scored
An 88 means the same thing on a product page, a clinic’s hours page, or a government form. Only the expected facts change by page type, never the points.
2xx on the agent path, robots policy, render independence
content negotiation, declared alternates, Vary, parity
core facts for this page kind, and what they cost in bytes
how deep into the response the first core fact appears
JSON-LD/microdata present, typed, complete, and agreeing with the text
canonical, cache validators, llms.txt, sitemap, declared endpoints
quantities, labelled facts and comparable rows an answer can quote
100
points
7
dimensions
63
measured
37
heuristic
63 of those 100 points are measured and 37 are heuristic. Every heuristic check is labelled wherever it appears, with the inputs it was inferred from.
Before you run one
ARS measures format and retrievability. It does not measure whether the facts are true, or whether any assistant cites this page.
Scanning is not publication. Nothing about a domain is listed, ranked, badged or indexed without verified owner opt-in, and removal is free, self-serve and permanent. The scanner identifies as rebilder-ars/0.2 and obeys robots.txt. Crawler policy, request budget and opt-out.
Questions
The answers below are the same entries that emit this page’s FAQ structured data, so an assistant reading the page gets exactly what a person reading it does.
An AI visibility checker fetches a page the way an AI assistant does and reports what came back. This one requests your URL as an agent and as a browser, compares the two responses, and scores the difference against the published Agent Readability Spec, so you can see which facts an assistant could reach and which it could not.
It is free and there is no account. Paste any public https URL and the result renders in the browser you ran it from.
One hundred points across seven dimensions: retrievability, machine representation, fact coverage, fact position, structured data, contract and discovery, and evidence density. The same dimensions apply to every kind of page; only the expected facts change. Measured checks and heuristic ones are labelled separately wherever they appear.
Yes, and the result page for an unclaimed domain stays noindex. No domain enters the index, a ranking or a badge without verified owner opt-in, and removal stays free, self-serve and permanent. Our own comparison pages are the one surface that prints a scan of a named site without its opt-in, and every number there carries its scan date, its ruleset and a link where you can run the same scan yourself.
It identifies as rebilder-ars/0.2, obeys robots.txt, keeps to a published request budget and never wears another crawler’s name. The full crawler policy and the opt-out are at /bots.
Every check shows the evidence it was scored from, so the result reads as a fix list rather than a grade, and the written fix for whichever check is costing you the most points is on the page for free. The rest of the written fixes unlock right on the result, as one page you can keep and share. The highest-leverage items are usually the same three: unblock the fetch, answer Accept: text/markdown with clean text, and move the facts an assistant is asked for to the front of the response.