AI Visibility Checker
See whether ChatGPT, Claude, Perplexity and Google can read your page. Real fetches, the evidence for every check, no signup.
Want a person to check the rest?
A script can read your HTML. It cannot tell whether your name, prices and service area match across your site, Google Business Profile and LinkedIn, whether Google and Bing have indexed the pages that matter, or what ChatGPT and Perplexity say when a customer asks. The free AI visibility audit checks those by hand: a one-page report within two working days, with the first three fixes in order.
Get the free auditAI assistants can only cite what their crawlers can read. This checker fetches your page the way they do: it reads your robots.txt rules for GPTBot, OAI-SearchBot, ChatGPT-User, ClaudeBot, Claude-User, PerplexityBot, Google-Extended, Googlebot and Bingbot, requests the page as Chrome and as each crawler, and compares what comes back. Then it looks at what a crawler that does not run JavaScript finds: the text, the title and H1, your structured data, canonical and noindex tags, and whether your email address survives Cloudflare's obfuscation.
Every result comes with its evidence: the robots.txt line, the status code, the byte count. What no checker can tell you is whether ChatGPT will mention you, and nobody can promise that. This removes the technical reasons an assistant would skip you. For the rest, the free AI visibility audit has a person compare your facts across the web and ask the assistants directly.
What the checker looks at
Crawlers allowed by robots.txt
Your robots.txt rules for OAI-SearchBot, ChatGPT-User, Claude-User, PerplexityBot, Googlebot and Bingbot, matched the way Google's own parser matches them, with the line that decides each one.
The same page for crawlers and browsers
We request the page as Chrome and as 8 crawlers and compare the status, size and title. Firewalls and "block AI bots" switches stop crawlers that robots.txt allows more often than you would think.
Text readable without JavaScript
The words, title and H1 in the raw HTML. Most AI crawlers do not run JavaScript, so a page built in the browser looks almost empty to them.
Structured data
The JSON-LD types on the page, any blocks that fail to parse, and the sameAs links that tie your site to your other profiles.
Open to search indexes
noindex, none, nosnippet and max-snippet:0 in meta tags and the X-Robots-Tag header. Assistants that search the web lean on search indexes.
Canonical URL
Whether the canonical tag points at the page itself or tells search engines to index another URL instead.
Email address readable
Cloudflare's email obfuscation turns addresses into "[email protected]" for anything that does not run JavaScript. We found it on our own site.
Training crawlers and llms.txt
Reported, never scored. Blocking GPTBot, ClaudeBot or Google-Extended is a business decision, and llms.txt is optional.
The crawlers it checks
Each AI company runs more than one crawler, and they do different jobs. Blocking the one that trains models is not the same as blocking the one that finds your pages for answers.
| Crawler | Used for | What it does | If you block it |
|---|---|---|---|
| GPTBotOpenAI | Used for Model training | Collects pages that may be used to train OpenAI models. | If you block it Your pages stay out of OpenAI model training. ChatGPT search is not affected: it uses OAI-SearchBot. |
| OAI-SearchBotOpenAI | Used for Search index | Indexes pages for ChatGPT search. | If you block it ChatGPT search does not show your pages in its answers. |
| ChatGPT-UserOpenAI | Used for Fetches for a user | Opens a page when a ChatGPT user asks it to. | If you block it ChatGPT cannot open the page for a user who asks. OpenAI says robots.txt may not apply to these visits; a firewall block does. |
| ClaudeBotAnthropic | Used for Model training | Collects pages that may be used to train Anthropic models. | If you block it Your pages stay out of Anthropic model training. |
| Claude-UserAnthropic | Used for Fetches for a user | Opens a page when someone asks Claude a question. | If you block it Claude cannot fetch the page for a user who asks about it. |
| PerplexityBotPerplexity | Used for Search index | Indexes pages so Perplexity can show and link them in answers. Perplexity says it is not used for model training. | If you block it Perplexity cannot show or link your pages in its answers. |
| Google-ExtendedGoogle | Used for Model training | A robots.txt setting, not a crawler: whether Gemini may use your pages for training and grounding. Googlebot does the fetching. | If you block it Gemini does not use your pages for training or grounding. Google Search, AI Overviews included, is not affected. |
| GooglebotGoogle | Used for Search index | Indexes pages for Google Search, AI Overviews included. | If you block it Google Search, and its AI Overviews with it, cannot read the page. |
| BingbotMicrosoft | Used for Search index | Indexes pages for Bing, which Microsoft Copilot also draws on. | If you block it Bing, and the assistants that draw on it, cannot read the page. |
User-agent strings as each company publishes them, checked October 2026. We send them from our own server; real crawlers also come from their company's IP addresses, which some firewalls verify, so a block in your results is strong evidence rather than proof.
Questions
What does an AI visibility checker check?
Whether AI assistants can find and read your website. This one tests the technical side: whether robots.txt lets their crawlers in, whether your firewall serves those crawlers the same page a browser gets, whether the text is in the HTML without JavaScript, and whether your structured data, canonical and noindex tags help or hurt. Some tools sold under the same name ask chatbots about your brand and count mentions instead; that is a different measurement.
Is this the same as an LLM visibility checker?
Yes. LLMs (large language models) are the technology behind ChatGPT, Claude, Gemini and Perplexity, and people use both names for the same question: can these assistants see my site?
If every check passes, will ChatGPT cite my site?
Not necessarily, and be careful with anyone who promises it. Passing means nothing technical is in the way. Whether an assistant names you depends on what it is asked, what other sites say about you and how plainly your pages answer the question. That is why the free AI visibility audit includes a spot check in ChatGPT, Perplexity and Google.
Should I block GPTBot?
That is a business decision, not a mistake. Blocking GPTBot keeps your pages out of OpenAI's model training, but ChatGPT search uses a different crawler, OAI-SearchBot. Anthropic splits the same way (ClaudeBot for training, Claude-User when someone asks), and so does Google (Google-Extended for Gemini, Googlebot for Search). The checker scores only the crawlers that find and open pages for answers.
robots.txt allows a crawler, so why was it blocked?
Firewalls act before robots.txt is ever read. Cloudflare's AI bot settings, Vercel's firewall and WordPress security plugins can turn a crawler away even when robots.txt lets it in. Some firewalls also check that a visitor claiming to be Googlebot really comes from Google, so a block on our Googlebot test can mean the firewall caught our imitation. Search Console's URL Inspection shows what the real Googlebot gets.
Do I need an llms.txt file?
No. We note whether you have one, but it never counts against you: support for it varies between assistants, while crawler access, readable HTML and consistent facts matter to all of them.
Do you keep the address I check?
Only briefly. Each result is cached for ten minutes so that repeating a check does not hit your site again, and the number of checks from your connection is kept for an hour to stop abuse. Nothing is stored after that, and we never use it to contact you.
Related
Want a person to check the rest?
The free AI visibility audit adds what a script cannot: your facts compared across your site, Google Business Profile and LinkedIn, indexing in Google and Bing, and a spot check in ChatGPT, Perplexity and Google. One-page report within two working days.