Find out if AI can read your website.

Paste your address. We'll show what our crawler received from your page, what your robots.txt says to the crawlers behind ChatGPT, Claude and Perplexity, and what the page's structure suggests about how readable it is: three kinds of evidence, each labelled.

Free, the full resultNo signupAbout 15 seconds

Scanning site... this usually takes 10-15 seconds.

AI-Readable Partially Readable Readable, few signals Not AI-Readable Inconclusive: page not observed (robots.txt, or no response)

Full report

Every check, with its evidence

Each check is labelled with the kind of evidence it rests on. None of them measures whether an assistant reads, understands, mentions, cites or ranks this page, or sends it traffic.

Fixes that apply

One item per check that fell short, derived only from the results above. Robots.txt items describe what the file declares; content and structure items are heuristic suggestions. A deliberate opt-out from AI training is never listed as something to undo.

Per-agent robots policy

Everything in this table is what your robots.txt declares, read from the file for this page's path. We identify honestly as CanAIReadMySiteBot and never send another crawler's name, so nothing here was tested by impersonating these agents, and none of it says what your edge does. For the edge, see crawl.access and crawl.declared_bot above: a block there is strong evidence about our request; a pass is weaker, since a verified vendor crawler from a published IP range may be treated differently than we are.

Generated robots.txt block

Names each AI retrieval agent (the crawlers behind ChatGPT, Claude and Perplexity) in its own User-agent block and repeats your site's own User-agent: * rules in each, so nothing else you restricted is opened. Where one of those rules blocks this page, that one rule is omitted for these agents and named in the comment: a deliberate change, stated, never a silent one. An agent your file already names keeps its own group. Training crawlers are left exactly as your site already has them. A robots.txt rule is a declaration: it does not change what your edge, WAF or CDN does. See the comment inside for details and exactly where to paste this in.

What AI sees

An excerpt of this page's own visible text, roughly as a text-based crawler would read it once scripts and markup are stripped away.

Generated llms.txt

Optional. Built from this page's own title, description, headings, and links -- ready to upload to your site root as-is. Worth 5 points in this model; not required for AI visibility, and major crawlers largely don't prioritise it.

Paste this into your site, then use Rescan now (in the actions under the score) to confirm it worked.

Embed your score

A small badge for your own site footer. Always regenerated from your latest logged scan -- never a stale number -- and links back to a live rescan here.

HTML

Markdown

Technical details

☕ This tool is free. If it saved you time, you can buy me a coffee.

The bit most tools get wrong

AI bots do two different jobs. Treat them differently.

Most tools treat every AI bot the same and tell you to let them all in. That isn't good advice: the two groups do completely different jobs.

Search crawlers: these can send you visitors

Retrieval agents

OAI-SearchBotChatGPT-UserClaude-UserClaude-SearchBotPerplexityBot

These fetch pages on behalf of an assistant or its search feature. They're the route by which a page can be cited or linked in an answer.

Blocking these tells AI search not to use your pages.

Training crawlers: these collect content for model development

Training crawlers

GPTBotClaudeBotCCBotApplebot-ExtendedMeta-ExternalAgent

These collect writing to train future models. By their operators' own accounts nothing comes back: no mention, no link, no visitor.

Blocking these is entirely your call.

We score them separately, so choosing to keep your work out of AI training never counts against you. More in the bot reference, or read the specifics for GPTBot, ClaudeBot and PerplexityBot individually.

What we keep seeing

The three most common problems we find

Each one is usually a setting that came bundled with your theme or host, not something anyone chose. All three are quick to fix.

Your robots.txt allows AI, but your host blocks it

Very common

Your robots.txt welcomes AI crawlers, but a security setting at your host turns away self-identifying crawlers like ours before they ever get that far. You'd never know without actually knocking on the door, which is what we do, as ourselves. We report what our request received; a verified vendor crawler may be treated differently.

Every page uses your company name as its heading

Common

Lots of themes put your business name at the top of every single page instead of that page's own title. So your article about spring delivery times reads, in its headings, like just another copy of your homepage.

Your navigation menu appears before your content, twice

Common

Many sites include their menu twice: once for phones, once for desktop. A text crawler reads both before it reaches a word about what you actually sell.

What we look at

What we check on every page

Nine scored checks and one reported-only access test, shown here in seven groups. Every one is explained in plain terms in your result, and labelled by the kind of evidence it rests on: what our request observed, what your robots.txt declares, or a heuristic reading of the page's structure. Full method, weights and ceilings on the rules page, including what we can't see.

crawl.access: What our self-identifying crawler actually received from your page, not just what your settings say should happen. Observed.
robots.retrieval: Whether your robots.txt allows the five retrieval agents that fetch pages for assistants. Declared. Reference
robots.training: Whether your training-crawler policy is internally consistent. Blocking training crawlers never costs points; one point is at stake for a same-operator gap only. Declared.
llms.txt: Optional. Whether you have one, and whether it follows the proposed structure. Five points; not required for AI visibility. What it is
content: Enough readable text and a clear headline, without needing JavaScript. Heuristic.
entity.signals: Whether the page says, in machine-readable terms, who you are and what you offer. Heuristic.
intent.trust: Whether your title, headline and description agree, and whether anyone can tell who publishes this. Heuristic.

Why it matters

Easy for people to read isn't the same as easy for AI.

More people now ask an assistant instead of searching. If a crawler can't reach or parse your page, an assistant has nothing of yours to draw on.
Plenty of good-looking sites are hard for a text crawler to parse. This tool makes that gap visible; it does not measure what any assistant does with your page.
Most problems are a single setting, not a rebuild. They're cheap to fix once you know which one.
We say who we are (CanAIReadMySiteBot) and never pretend to be someone else's bot. About our crawler.