Can AI search engines
actually read your site?
Enter your homepage. We check whether the crawlers behind ChatGPT, Claude, Perplexity, Google AI Overviews and Copilot are allowed to fetch it, whether your server actually serves them, and whether there is anything readable there when they arrive.
Takes about five seconds · We identify ourselves honestly as ProWebsiteAudits-AIReadiness, we never pretend to be a real AI crawler
Four verdicts, no score out of 100
A colour-coded number tells you nothing you can act on, so we do not produce one. Every check comes back as one of four things, with the reason and the fix written out in full.
Seventeen crawlers, two very different jobs
Most free checkers lump every AI bot into one list and tell you off for blocking any of them. That is bad advice. Half of these decide whether an assistant can cite you today; the other half only collect training data, and turning those away costs you nothing.
Crawlers that decide whether you get cited
Blocking any of these quietly removes you from the answers people are already reading.
OAI-SearchBotOpenAIBuilds the index behind ChatGPT Search.ChatGPT-UserOpenAIFetches your page live when someone asks ChatGPT about you.Claude-SearchBotAnthropicIndexes pages so Claude can surface and cite them.Claude-UserAnthropicFetches your page when a Claude user asks about it.PerplexityBotPerplexityBuilds the Perplexity index.Perplexity-UserPerplexityFetches your page when someone follows a Perplexity citation.GooglebotGoogleAI Overviews and AI Mode are built on the normal Google index.BingbotMicrosoftCopilot answers are grounded in the Bing index.ApplebotApplePowers Siri and Spotlight suggestions.Crawlers that only feed model training
Blocking these is a content-licensing decision. It has no effect on whether you are cited, so we never mark it as a failure.
GPTBotOpenAICollects pages for training OpenAI models.ClaudeBotAnthropicCollects pages for training Anthropic models.Google-ExtendedGoogleControls use of your content for Gemini training and grounding.Applebot-ExtendedAppleControls use of your content for Apple Intelligence training.CCBotCommon CrawlFeeds a large share of open model training sets.meta-externalagentMetaCollects pages for Meta AI.AmazonbotAmazonFeeds Alexa and Amazon shopping assistants.BytespiderByteDanceCollects pages for ByteDance models.What this tool cannot tell you
It checks one page for access and readability. That is a real question with a real answer, and it is worth knowing. But being reachable is not the same as being cited, and no automated check of your own HTML can close that gap:
Common questions
Seventeen, split into the ones that decide whether you can be cited today (OAI-SearchBot, ChatGPT-User, Claude-SearchBot, Claude-User, PerplexityBot, Perplexity-User, Googlebot, Bingbot, Applebot) and the ones that only collect content for model training (GPTBot, ClaudeBot, Google-Extended, Applebot-Extended, CCBot, meta-externalagent, Amazonbot, Bytespider). We treat those two groups very differently, because blocking the second group is a legitimate choice that costs you nothing in AI answers.
No, and this is the single most common misunderstanding. GPTBot collects pages for training OpenAI models. OAI-SearchBot builds the index ChatGPT actually searches, and ChatGPT-User is what fetches your page when someone asks about you. You can block GPTBot to keep your content out of training and still be cited every day, as long as the other two are allowed.
Googlebot renders JavaScript. Most of the AI crawlers do not. If your homepage ships an empty shell and paints the content in after load, GPTBot, ClaudeBot and PerplexityBot see a blank page, no matter how good the copy is. That is why we count the words present in the raw HTML before any script runs.
No. This checks whether AI crawlers can reach and read one page. It cannot tell you whether ChatGPT actually names you when someone asks about your category, how often you are cited against named competitors, or which sources those answers are drawing on. That is what the paid AI Visibility Audit does.
We log the domain and the outcome so we can spot bugs and keep the tool honest. Nothing is emailed, nothing is shared, and there is no form to fill in before you see your result.
It is a proposed convention for a plain-text file describing your site to language models, similar in spirit to robots.txt. We check for it and report it, but none of the major assistants currently honour it. It costs very little to add and it is not going to move anything on its own yet.