Is your website invisible to AI? How to check if ChatGPT can even read it

The most expensive mistake in AI visibility is not weak content or a low score. It is having a website the AI cannot read at all. You can publish the clearest, most helpful page in your industry, but if ChatGPT’s crawler cannot reach it, none of it counts.

The good news is that this is the most fixable problem in the whole field. Being invisible to AI is almost always an accident, a setting somebody flipped or a default nobody noticed, and accidents can be undone. Here is how to tell whether your site is readable, and what to do if it is not.

Being readable comes before everything else

Before an AI can recommend you, it has to be able to read you, and that is the first of three gates every page must clear. For a model to ground an answer in your content, a bot has to reach the page, the model has to extract a clear fact from it, and that fact has to surface for the question being asked. Those gates are crawlable, parseable, and retrievable, in that order.

Crawlable comes first because nothing downstream works without it. A generated answer usually pulls from somewhere between five and sixteen sources, and a page a crawler never reached cannot be one of them. If you remember one thing, remember this: if you cannot be read, nothing else you do for AI visibility matters.

Can ChatGPT actually read your website?

Not always, and that is the uncomfortable part. AI assistants do not browse the web the way a person does. They rely on automated crawlers with specific names, and your site quietly decides which of them it lets in. The main ones to know are:

• GPTBot, the crawler OpenAI uses to gather web content for ChatGPT.
• ClaudeBot, Anthropic’s crawler for Claude.
• PerplexityBot, which feeds the Perplexity answer engine.
• Google-Extended, the control that decides whether Google may use your content in AI features like Gemini.

Each of these checks a small file on your site called robots.txt before it reads anything. That file is a short list of instructions telling crawlers where they may and may not go. It is powerful, and that is exactly why it is so easy to get wrong.

The most common way businesses block themselves

The single most common and most embarrassing GEO failure is a site that accidentally blocks the AI crawlers in its own robots.txt file. A developer adds a rule to keep bots out, a template ships with AI crawlers disallowed by default, or a security tool tightens things up, and suddenly GPTBot and its peers are turned away at the door.

This is not a fringe problem. Cloudflare, which sits in front of a large share of the web, at one point set its defaults toward blocking AI crawlers, so a site could shut the door without anyone choosing to. And the traffic at stake is real. BrightEdge has reported that AI agents and crawlers already make up roughly a third of organic crawl and search activity, so turning them away is no longer a rounding error.

When your own design hides you

Even with a welcoming robots.txt, your site can still be unreadable because of how it is built. If your important content only appears after heavy client-side JavaScript runs, or sits behind a login or a paywall, a crawler often sees an empty shell where your words should be.

The test is simple in principle. If a fact about your business cannot be seen in the plain page that loads before any scripts run or any sign-in, assume an AI cannot see it either. Your best selling points belong in readable text on a public page, not locked inside an app-like experience only humans with accounts can open.

How to check whether you are invisible

You can check the basics yourself in a few minutes, before you spend a dollar on content. Start with the plumbing and work outward:

• Open yoursite.com/robots.txt and look for any line that disallows GPTBot, ClaudeBot, PerplexityBot, or Google-Extended. If they are blocked, that is your problem.
• Check your CDN or security provider, such as Cloudflare, for a bot-blocking setting that may be switched on by default.
• Load a key page with JavaScript turned off, or view its source, and confirm your main facts are actually in the text.
• Make sure nothing a customer needs to find you is trapped behind a login or paywall.

If all four look clean, the door is open and you can move on to making your content easy to quote. If any of them is off, you have likely found the cheapest visibility win you will ever make.

Does adding an llms.txt file fix this?

No, and it is worth being clear about that because it is widely oversold. An llms.txt file is a proposed, robots.txt-style map that points AI systems to your most important content, and around 844,000 sites had adopted it by early 2026. It takes about thirty minutes to add and it does no harm.

But the honest verdict from technical SEO analyst Kevin Indig is that it is "a good idea that lacks confirmed impact," worth adding because it is low-cost, not because it is proven. It is not a substitute for being crawlable in the first place. Fix the door before you put up a nicer sign.

See whether the AI can find you

The fastest way to find out whether AI can read and recommend your business is to look at the answers themselves. Crawlability is the gate, but what you really want to know is whether you make it into the short list of names the AI gives when a customer asks about your category.

That is exactly what PromptRank measures. It runs real buyer-intent questions across ChatGPT, Claude, Gemini, Grok, and DeepSeek, scores your AI visibility from 0 to 100, and shows you the competitors winning the answers you are missing. Run a free audit at trypromptrank.com and find out whether the AI can see you at all.