See who reads your siteEven the AI
Google Analytics counts people. It doesn't count ChatGPT, Claude or Perplexity reading your pages, and it can't tell you which of those visitors are fake. Your hatamoto.io site records all of it and shows it in plain language.
Free for 14 days. Card needed, nothing charged until the trial ends.

Every assistantEvery page
Your site keeps a record each time a known AI crawler or assistant fetches a page: who it was, which page it read, and whether your site answered properly. Hatamoto knows 43 of them by name, from GPTBot and ClaudeBot to PerplexityBot and Googlebot.
- Which pages get read. Your most fetched pages, ranked, with each assistant's own activity.
- How fast you were found. See when an AI first read a new page, and how long after you published it.
- Nothing to install. Your server records the visits itself, so there is no tracking script to add and no plugin to update.
The visitors AI sends you
When a visitor clicks a link to your site inside an AI assistant, they arrive as their own channel. You see how many came, from which assistant, and what share of your traffic that is.
- Nine assistants tracked: ChatGPT, Perplexity, Claude, Gemini, Copilot, Grok, Meta AI, DeepSeek and Mistral.
- No cookie, no visitor ID. Only the assistant's name and the day are stored. Not the person, not their address.
- Trace it to a page. Each post shows how often assistants fetched it in the last 7 and 30 days.
Visits sent by AI assistants
Harbour Street Bakery · last 30 days
Example numbers. AI share of all visits: 17%
Verification
Where each visit actually came from
Catch the bots in costume
Anyone can type "GPTBot" as their name. So your site checks every visit against the addresses each company publishes, and against reverse DNS where a company offers it. Visits that fail the check are marked as fakes, and you can read the list.
- Verified, Fake or Can't check. Every crawler visit carries one of those labels, so your numbers are honest.
- Fakes stay out of your page numbers. A forged name can't make a page look like it was read.
- A health banner. If too many visits fail, the admin warns you that something is off with how visitors are read.
Technical detail: ranges come from Google, Bing, OpenAI, Perplexity, Anthropic, CCBot, DuckDuckGo, Apple and Ahrefs, downloaded by a background job and kept in your site's database.
Allow or blockBy purpose
Not every AI visit is the same. A company training a model on your writing is different from an assistant fetching your opening hours because a customer asked. You choose, one purpose at a time, and the site updates its robots.txt for you.
- Training, AI search, user fetches. Plus search engines, social previews, SEO tools and web archives.
- One switch each. No file editing. Only the owner can change the policy.
- Honest about limits. robots.txt is a request, not a lock. The admin marks the crawlers that are known to ignore it.

The gapsIn a list
Two reports tell you where to look first.
Published pages no AI has read
Link to these from pages that get read
/classes/kids0 fetches/blog/rye-starter0 fetches/catering0 fetches
Errors served to AI crawlers
Old addresses they keep asking for
/menu-2022404/shop/loaves404/old-classes404
Choose Create redirect to send them to the right page.
One month on a real siteMostly invisible
These are the last 30 days on hatamoto.dev, the site for the open-source software your hatamoto.io site runs on.
Plus 863 published pages that no AI had fetched yet. Numbers from hatamoto.dev, last 30 days.
All of itBuilt in
43 named agents
Each sorted by purpose: training, AI search, user fetch, classic search, social, SEO and archive.
First crawls
Which page an AI read first, which agent it was, and how long after you published.
Per-page view
An AI crawlers card on every post: fetches over 7 and 30 days, and each agent's last visit.
Google's view too
Googlebot crawls, pages it has not visited, errors and stale pages, from your own site's records.
Any date range
Up to a year at a time. Daily totals are kept, so older reports stay complete.
Private by design
No cookies, no visitor IDs. After 30 days a crawler's address is cut down to its network. Nothing leaves your site.
Crawler visits are recorded by the server, so they need no script. People are counted by a small cookieless page-view beacon that the default theme sends. AI reports cover pages, robots.txt, sitemaps, llms.txt and feeds.

Right next to everything else
AI traffic sits on your admin overview with your pages, drafts, form enquiries and site health. No second dashboard, no separate login.
- Every plan gets an AI visibility report.
- Business adds full AI analytics with fake-bot blocking.
QuestionsAnything else, just ask
Talk to usWe can't promise that, and nobody honest can. It shows you what assistants read and where they hit errors, so you can fix what's in your control.
You can keep it. Your site also counts pageviews and AI-referred visits without cookies, so you can see the AI side that Google Analytics misses.
For crawlers that follow the rules, yes. Your robots.txt asks them to stay away. Some, mostly those that fetch a page because a person asked, can ignore it. The admin marks them.
Some agents publish no address list, so there is nothing to compare against. We show that honestly instead of calling it real.
For people, only a daily pageview total and a daily count per AI assistant. No cookie, no visitor ID, no address.
Yes. With a key you control, it can read crawler and referral numbers through the MCP server.
Find out who's reading you
Start your free 14-day trial. Card needed, nothing charged until the trial ends.