Can AI engines read your website?
Enter your domain. Eleven checks show what ChatGPT, Perplexity and Google's crawlers receive from your homepage, and what shuts them out.
What the checker looks at
Each check copies what a crawler does on its first visit to your site. Problems that stop an engine outright weigh most in the score.
01
Homepage loads
Your homepage answers with status 200 and real HTML, not an error page or a firewall challenge.
02
Open to indexing
No noindex tag or header asks search and answer engines to drop the page.
03
AI answer bots allowed
robots.txt, the file that tells bots where to go, lets in the six bots that read pages live.
04
AI training bots
Training bots such as GPTBot and ClaudeBot. Blocking them is a fair choice, so it only warns.
05
Google and Bing allowed
Googlebot and Bingbot can crawl. AI Overviews use Google's crawl, and Copilot draws on Bing's index.
06
Text without JavaScript
At least 150 words arrive in the HTML itself, because many AI crawlers never run JavaScript.
07
Structured data
JSON-LD structured data, the machine-readable summary of a page, is present and parses cleanly.
08
Named owner in markup
Your markup names an organization or person, with sameAs links to profiles that confirm who you are.
09
Page basics
One title, one meta description, a canonical link and exactly one H1 heading.
10
llms.txt file
A /llms.txt file exists. No engine has confirmed reading one, so this weighs least.
11
Sitemap
robots.txt declares a sitemap, or /sitemap.xml lists your pages.
The bots it tests
Answer bots
Fetch pages live to cite them. Blocking one fails.
- OAI-SearchBot (OpenAI)
- ChatGPT-User (OpenAI)
- Claude-SearchBot (Anthropic)
- Claude-User (Anthropic)
- PerplexityBot (Perplexity)
- Perplexity-User (Perplexity)
Training bots
Gather model training data. Blocking one only warns.
- GPTBot (OpenAI)
- ClaudeBot (Anthropic)
- Google-Extended (Google)
- Applebot-Extended (Apple)
- CCBot (Common Crawl)
- Meta-ExternalAgent (Meta)
Search bots
Feed Google and Bing, which AI answers draw on.
- Googlebot (Google)
- Bingbot (Microsoft)
What it cannot tell you
The checker reads your site the way a crawler would on one visit. Four things sit outside that view.
Whether any engine cites you
Readable is not cited. Only asking ChatGPT and Perplexity real questions shows whether they name you.
Blocks aimed at verified bots
Some firewalls block only verified GPTBot or ClaudeBot addresses. The checker sends from neither, so those blocks stay hidden.
Whether an engine chooses to search
An assistant can answer from memory without searching the web. No setting on your site changes that.
Pages past your homepage
It reads the homepage, robots.txt, llms.txt and the sitemap. Deeper pages can differ.
The fixed-scope SEO and answer engine optimization (AEO) audit asks live ChatGPT and Perplexity your buyers' questions, and records whether they name you. Generative engine optimization (GEO) is the work that raises how often they do. The SEO and AEO best practices reference explains the crawler rules behind each check.
AI visibility, answered
Short answers on each check and the blind spots of the score.
The AI visibility checker tests whether AI crawlers can reach and read a site's homepage, across eleven checks. It fetches the homepage, robots.txt, llms.txt and the sitemap, the way a crawler does on its first visit. It then checks which AI answer, training and search bots robots.txt admits. It also checks whether the page's text arrives without JavaScript, and whether its structured data names the owner. Pooya Golchian built it as a free first pass before the paid SEO and AEO audit.
No. Blocking GPTBot keeps a site out of OpenAI's model training, while ChatGPT search finds pages through OAI-SearchBot. ChatGPT-User fetches a page when a user asks ChatGPT to read it. OpenAI states that a site can block GPTBot and still appear in ChatGPT search while OAI-SearchBot is allowed. Anthropic splits the same way, with ClaudeBot for training and Claude-SearchBot and Claude-User for retrieval. The checker fails a blocked retrieval bot and only warns on a blocked training bot.
Many AI crawlers read only the HTML a server sends and never run JavaScript, so text drawn in the browser does not exist for them. A client-rendered single-page app can look complete to a visitor and empty to GPTBot or PerplexityBot. Googlebot does render JavaScript, which is why a site can rank on Google and still go uncited by ChatGPT. The checker counts the words present before any script runs and fails a homepage under 150.
No engine has publicly confirmed that it reads llms.txt, so the checker gives the file the lowest weight of all eleven checks. llms.txt is a plain text file at the site root that lists a site's key pages for AI agents. Pooya Golchian treats it as cheap insurance rather than a ranking lever. A missing file only warns, and barely moves the score.
The checker is free, needs no email, and does not store the address you enter. It fetches four public files from that site, builds the report and sends it to your browser. It keeps a count per IP address for one hour to cap use at ten checks, and the count then expires. It names itself honestly to the site it checks, as PooyaGolchianAIVisibilityChecker, and never poses as GPTBot or any other crawler.
Engagement
Next, measure whether engines cite you
Size the work on a free call before you commit to anything.
The SEO and AEO audit is $6,000 fixed.
About AED 22,000.
One fixed-scope pass over technical SEO, the entity graph and measured AI citation rate, in 2 to 3 weeks. You get a findings document, a ranked fix list with file-level locations, and a re-measure at 60 days. The fee is credited toward a Generative Engine Optimization engagement booked within 60 days.