How it works
Four steps from an AI visit to a plain answer.
Connect your site in two minutes.
Every visitor is checked three ways.
- 1
Its name
The user agent says GPTBot, ClaudeBot or Googlebot. Anyone can type that, so it's only the start.
- 2
Its address
The IP must be in the ranges the company publishes. OpenAI, Google, Perplexity and others list theirs.
- 3
Its reverse DNS
The address must resolve back to the company's domain, and forward again. Where agents sign requests, we check the signature too.
Then sorted by why it came.
Fetched right now because someone asked an AI. The closest thing to a real question.
Builds the index assistants search before they answer.
Collects pages that may be used to train future models.
Browses and acts on your site on behalf of a person.
Classic search engine crawlers, for comparison.
SEO tools, link previews and monitors. Counted, kept out of the way.
And turned into answers you can act on.
See which AI reads which page.
Count the people AI sends you.
See what people probably asked.
Find the pages AI can't open.
Catch the fakes pretending to be Googlebot.
Read it all in plain English, every Monday.
ChatGPT opened /services/implants and /prices together 63 times.
People are asking ChatGPT how much implants cost nearby.
Add a clear “from” price so ChatGPT can quote it.
Measured is measured. Estimates say so.
- Which bot visited, and whether it was real
- Which pages, when, and how often it came back
- Errors and redirects it hit
- People who arrived from an AI answer
- The likely topic behind a group of visits
- What a change in visits probably means
- What to do next