Can AI find your
clients’ websites?
AI engines don’t rank pages. They cite authorities.
CiteCore queries five AI engines by name and tells you exactly where your brand appears — and where it doesn’t.
The Shift
Search gave you 10 blue links. AI gives you one answer.
There is no page two in an AI response. Your clients are either the cited authority or they are invisible. CiteCore tells you which one — and fixes it.
Universal AI Visibility Matrix
This is what an audit looks like.
4/5 engines citing
ChatGPT
OpenAI
“Best fractional CIO services for mid-market companies”
Claude
Anthropic
“Which firms handle AI visibility for agencies”
Gemini
“AI search optimization tools for agencies”
Perplexity
AI Search
“Top-rated AI visibility optimization platforms”
Grok
xAI
Illustration, not a live scan. The five engines shown are the five CiteCore queries; the detection states and confidence numbers are placeholders. Your audit reports what these engines actually returned.
Scoring Methodology
Five files. 20 points each.
100 total. AEO measured.
The Core AEO Scan grades the five files that decide whether an AI engine can read a site at all. Scoring runs in code, not in a language model, so the same site always produces the same score.
robots.txt
Which named AI crawlers are permitted, which are blocked, and which are blocked by a default the owner never set
llms.txt
A structured summary at a fixed path — present, resolving, one top-level heading, describing the same entity as the site
llms-full.txt
The long-form companion: sectioned, single H1, and consistent with llms.txt rather than contradicting it
sitemap.xml
Every advertised URL resolves, every page is advertised, and lastmod is present so a crawler can schedule a re-read
Structured data
The organization identified once sitewide, and each page declaring its own topic instead of repeating one block
What You Receive
An audit reports what the engines returned, not what they might
CiteCore sends a real question to each of the five engines and records the answer. Everything in the report traces back to one of those recorded answers or to a file read from the live site.
Which engines named the business, and in what words
For each of ChatGPT, Claude, Gemini, Perplexity and Grok, the audit records whether the brand was named in the answer and reproduces the passage that named it. An engine that returned nothing is reported as returning nothing — CiteCore does not infer a citation from a competitor's absence.
The Core AEO score, with the evidence behind each of the five files
Every one of the five control files is graded out of 20 and the report shows what was read at the live URL. A score is never a summary of a file CiteCore could not fetch: an unreadable file leaves the calculation and the report names it.
The specific edits that move each file's score
A CiteCore finding names the file, the line, and the change. "Add structured data" is not a finding. "Your sitemap advertises four URLs and seven pages resolve; three pages are invisible to any crawler that trusts it" is a finding, because someone can act on it without a second conversation.
Which AI crawlers your robots.txt currently blocks
Including the ones blocked by a managed default the site owner never chose. A single disallow line aimed at GPTBot, ClaudeBot or PerplexityBot removes a site from those engines' answer sets entirely, and it is the most common finding CiteCore reports on a site that believes it is optimized.
The Five Engines
Each engine decides differently, so each is asked separately
Being cited by one is not evidence of being cited by another. CiteCore records each engine's answer on its own rather than averaging them into a single visibility number.
ChatGPT
Reaches the open web through its own crawler, GPTBot, and through a separate agent used when a user asks it to browse. A site can permit one and block the other without realising it, which is why CiteCore reports the named agents in robots.txt individually rather than reporting a single allow-or-block verdict.
Claude
Fetches with ClaudeBot, and weights clearly structured, plainly written source material. Sites whose substance is delivered by client-side rendering are read as thinner than they look in a browser, because what the crawler receives is the shell rather than the text.
Gemini
Draws on Google's index and on structured data already published for search. This is the engine where existing SEO investment carries across most directly — and also where an entity described inconsistently across pages does the most damage, because the inconsistency is visible in markup Google already stores.
Perplexity
Cites sources inline and links them, which makes its answers the easiest to verify and the fastest to change. PerplexityBot is separately named in robots.txt, so a site can be readable to every other engine and invisible to this one on the strength of a single line.
Grok
Runs on xAI's models with access to real-time content. Answers shift faster here than on engines working from a periodically refreshed index, so a single reading is a snapshot rather than a settled position, and CiteCore reports it as one.
Ready to see where your brand stands?
Request a complimentary AI Visibility Audit and discover exactly how AI engines see your business today.
Request Your Audit