Tools › AI Presence Site Check
AI Presence Site Check
This check takes a short agent walk through your public site and searches for lists and profiles. It tries one answer engine and falls back to a web-search sample. You get the route, the evidence, and evaluator notes. That is one sample, not a score.
Task
The agent's path
Lime fetched · outline missed · amber blocked, skipped, or a name collision
What the web returned
Evaluator notes
Off-site evidence
Owned-site checks
Supporting evidence from the pages in this walk.
This search is a sample. The first month maps the rest.
A few public hits are not the category map. Month one checks the lists, answers, and profiles this run can miss, then starts work on the gaps.
Book a call →What this check looks at
The agent starts at the URL you paste, follows a capped set of same-site links, and records what each fetch returned. It checks up to five HTML pages of 512 KB each, plus robots.txt, llms.txt when it exists, a public-web search, and one entity-recall sample. It does not run your site's JavaScript.
- Name and canonical URL
- A plain description of the product
- Public pricing, or how pricing is explained
- Named customer proof
- Buyer-facing pages we can fetch from those links
- robots.txt as a file we fetched, not as an AI-access grade
- Third-party lists and profiles in a public web search
- One answer-engine reply, or one web-search sample when no answer engine is configured
The map shows this walk, not every page on the site. One reply or sample is not a ranking, citation, score, or the retainer map.
FAQ
The check acts like a lightweight web agent. It opens your homepage, follows a capped set of public links, checks robots.txt and llms.txt when available, and searches for third-party lists and profiles. It tries one answer engine and falls back to a sample synthesized from up to five web-search results. It reads returned HTML and does not run your site's JavaScript.
No. The check tries one answer engine. If that is unavailable, it shows a web-search sample of what an answer would see. One reply or sample is not a ranking, citation, or score.
The walk and search are capped. A missing node or check means this run did not find evidence in the pages it fetched or the search sample. It does not prove the page or profile does not exist.
Some hosts put a Cloudflare-style challenge in front of the HTML. This check does not run that JavaScript, so it cannot read the owned pages. A public web search may still run. That is the site blocking automated fetches, not a ranking result.