How do I check if AI can see my website?

Check two things. Whether your robots.txt permits each AI user-agent token individually — GPTBot, OAI-SearchBot, ChatGPT-User, ClaudeBot, Google-Extended and PerplexityBot are governed separately, and allowing Googlebot grants none of them. And whether your content exists in the HTML your server returns, since AI crawlers generally do not execute JavaScript.

Why is my site not appearing in AI answers?

Most often one of two causes. A robots.txt rule blocking the relevant token, frequently a blanket AI block added years ago and forgotten. Or content that only exists after client-side rendering, leaving the crawler an empty page. Both fail silently — search rankings are unaffected, so nothing signals the problem.

Is there a free AI visibility checker?

Yes. Ours is free and needs no account for your first scan. It tests crawler access per user-agent token, answer-engine extractability, and whether an automated agent could operate the page — three separate capabilities rather than one blended score.

AI Visibility27 July 20267 min read

Free AI visibility checker: three capabilities, one scan

Whether AI assistants can reach your site, quote it accurately and operate it are three separate questions. Most sites pass one and fail the others, with no symptom either way.

By Sapun Lamichhane · Arcetis

AI visibility is not one thing, and treating it as one is why so much advice about it is useless. It is three capabilities that fail independently, and a site routinely has one in perfect order while another is completely broken.

Can they reach you?

Governed entirely by robots.txt, per user-agent token. Every operator crawls under its own name and each is controlled separately — OpenAI alone runs three, doing three different jobs.

The mechanic that catches people is that robots.txt groups are not additive. A crawler obeys the single most specific group matching its name and ignores every other, including the wildcard. So a permissive group added for one AI crawler does not give it your site-wide rules plus permission — it releases that crawler from your site-wide rules entirely.

Can they quote you?

Reachable is not quotable. An assistant has to locate an answer, bound it, and decide it survives being lifted out of its context. Prose that builds toward a conclusion across several paragraphs is prose it will paraphrase loosely or skip.

What makes a passage liftable is structural: a heading phrased as the question, the answer stated immediately beneath it, semantic landmarks separating content from navigation, and a named author with a visible date to attribute it to.

Can an agent operate you?

The newest question, and the one nothing else measures. Software that books and buys on a person's behalf does not see your page — it reads the accessibility tree.

So a div styled as a button is not a button. An input with a placeholder and no label has no name to match against. A hover-only menu never opens, because there is no pointer. None of this affects your ranking, and all of it decides whether an agent completes a task on your site or reports that it could not.

Why all three fail silently

This is the part worth internalising. None of these failures produce an error, a warning, or a change in any metric you already watch. Your rankings hold, your analytics look normal, and you are simply never mentioned. The only way to know is to test the inputs directly.

What free means here

Our AI visibility checker is free, with no account needed for your first scan. It reports the three capabilities separately rather than averaging them, because an average hides exactly the failure you need to find.