Five short checks answer it without an account: can AI crawlers reach your pages, is your llms.txt valid, does JavaScript hide your text, does your schema name you clearly, and is a bot in your logs real. Each one runs in under 20 seconds and hands you the evidence, not a score.
Each tool isolates one question that also sits inside the full Reflexa page check's 23 checks: can AI crawlers reach the page, is llms.txt valid, is the text still there once JavaScript is stripped out, does structured data name the company clearly, and is a bot that visited your logs genuine. They are a free, cut-down slice of everything the Reflexa agent does on your site.
| Tool | Answers | Time | A bad result looks like |
|---|---|---|---|
| AI Crawler Access Check | Can ChatGPT, Claude and Perplexity read the site? | ~5 seconds | One or more crawlers disallowed in robots.txt, or allowed there but blocked by a CDN |
| llms.txt validator & generator | Does llms.txt exist and follow the spec? | ~5 seconds | Missing file, broken links, or sections out of the format the spec expects |
| What AI actually sees | Is the text there without JavaScript? | 5 to 20 seconds | Pricing, FAQ or product copy that only appears after a script runs |
| Schema & entity check | Does structured data name the company clearly? | ~3 seconds | Organization schema missing, or naming a different entity than the visible page copy |
| AI bot verifier | Is a bot in the logs really GPTBot, ClaudeBot or PerplexityBot? | Instant | A user agent claiming to be a known crawler from an IP outside that vendor's published range |
Only if your robots.txt allows each crawler's user agent, and nothing in front of your site, like a CDN or bot wall, blocks it anyway. /tools/ai-crawler-check reads your robots.txt rules for eight AI user agents, including GPTBot, OAI-SearchBot, ClaudeBot and PerplexityBot, then makes a live fetch with each bot's real user agent to catch CDN blocks that robots.txt alone never shows.
What a bad result looks like: a crawler disallowed outright in robots.txt, or one that robots.txt allows but a CDN or bot-protection rule still turns away with a 403 or a challenge page.
llms.txt is a proposed Markdown file, defined at llmstxt.org, listing your key pages so an AI reader gets a clean map of your site instead of guessing from your sitemap. /tools/llms-txt checks an existing file against that spec and tries every link in it, or drafts one from your sitemap and page titles if you don't have one yet.
What a bad result looks like: no file at all, a file with links that 404, or sections that don't follow the format the spec expects.
Only a real render without scripts can tell you, which is what /tools/what-ai-sees does. It fetches your page as raw HTML and as a headless render, and shows the two side by side, so you can see how much text a crawler that skips JavaScript never gets to read.
What a bad result looks like: a hero section, pricing table or FAQ block that only appears after a script runs, meaning any crawler that skips JavaScript sees a mostly blank page.
An entity, the one distinct thing AI treats as "this company," has to be named the same way in your structured data and your visible copy, or AI is left guessing which one is real. /tools/schema-entity runs thirteen checks on your homepage's JSON-LD: Organization data, a name that mirrors your visible copy, and sameAs links that point at real profiles.
What a bad result looks like: no Organization schema at all, a legal name in the schema that doesn't match the name on the page, or sameAs links that 404.
By checking its IP against the vendor's own published ranges, which is what /tools/bot-verifier does. Paste an IP and user agent, or a whole log line, and it tells you whether the visit matches OpenAI's, Anthropic's or Perplexity's real crawler infrastructure, or a scraper spoofing a known user agent.
What a bad result looks like: a request claiming to be GPTBot or ClaudeBot from an IP address outside that vendor's published range, which usually means a different bot copied the user-agent string.
Crawler access first, because nothing downstream matters if the engine can't reach the page at all. After that: JavaScript, then schema, then llms.txt, and check your logs for bot authenticity only once you suspect a specific visit. That order matches how the full page check groups its own findings, covered in how to read a Reflexa page check.
None of the steps below require buying anything:
Bottom line: the five free tools answer whether AI can read and understand your site; only the full check answers whether it recommends you.
Yes. Each one runs with no account, no signup and nothing stored beyond a short cache for shareable links. They are cut-down versions of the checks the paid Reflexa page check runs, each reduced to a single question.
No. They check whether AI can reach, read and correctly identify your site, which is a precondition for being recommended, not the same thing. Whether they recommend you is decided by what you say and who else writes about you; the full page check at dashboard.reflexatechnologies.com/audit covers the readable half on every page.
AI Crawler Access Check. If a crawler is blocked at the robots.txt level or by a CDN, nothing else about your content matters yet, since the engine never gets far enough to read it.
Most return a result in about 5 seconds. What AI Actually Sees takes longer, 5 to 20 seconds, because it runs a real headless render of your page rather than just fetching the raw HTML.
Each tool fetches a handful of public pages from your own site, the way a crawler would, computes the result, and caches it briefly so a shared link loads fast. No account is created and nothing is stored beyond that cache.
The free check reads your pages the way ChatGPT, Claude and Perplexity read them, 23 structural checks, evidence attached, no card needed.