The AI card sits below the geography and technology cards on your dashboard.
The AI card on the dashboard, populated, with the Crawlers tab selected and a realistic distribution across operators (e.g. OpenAI 612, Anthropic 341, Google 188, Perplexity 74). Light theme, desktop 1440px.
Three populations, kept apart
Most analytics products sort traffic into two buckets: people, and everything else they throw away. TrueStat keeps three.
| What it is | Where it shows | |
| Humans | People with browsers | The main tiles and every breakdown |
| AI crawlers | Assistants and training crawlers fetching your pages | This panel |
| Search crawlers | Googlebot, bingbot and the rest, indexing for search | This panel, in its own category |
Keeping AI crawling separate from search crawling is the point. They are different operators doing different things for different reasons, and folding Googlebot in with GoogleOther — as a simple "is this a bot" flag does — makes the number meaningless.
The tabs
Crawlers
Views per operator — OpenAI, Anthropic, Perplexity, Google, Microsoft Bing, Apple, Meta, ByteDance, Common Crawl, Other.
This answers "who is reading me".
Pages read
Which of your paths the crawlers fetched, ranked. Shows the top 12. See AI traffic by page.
Crawler types
Views split by why the crawler came, rather than by who sent it:
| Type | Meaning | Examples |
| AI assistant | Fetching to answer a person's question right now | ChatGPT-User, Claude-User, Perplexity-User, OAI-SearchBot |
| AI training | Collecting a corpus for model training | GPTBot, ClaudeBot, GoogleOther, Bytespider, CCBot |
| Search indexing | Ordinary search engine indexing | Googlebot, bingbot, Applebot |
| Other bots | Uptime monitors, scrapers, anything self-identifying |
The assistant/training split is the one worth watching. An AI assistant fetch means someone asked a question and your page was pulled in to answer it — that is a reader, one step removed. An AI training fetch means your content is being collected for a future model, with no reader at the other end today. They have opposite implications and you probably feel differently about each.
This tab also shows how much of each category was IP-verified. See How reliable is it.
Machine-readable exports
Fetches of the surfaces built for machines rather than browsers — llms.txt,
llms-full.txt, a page's .md variant, an MCP endpoint. If you publish any of
these, this is where their traffic appears.
If you publish none, this tab is empty and that is expected.
The two dashboard tiles
Above all of this, on the main tile row:
AI crawler views — total crawler page views in the range. It ignores the "Hide bots" filter, because it exists to report what that filter removes.
Verified crawlers — the share confirmed by IP range rather than merely
claimed. A dash means no crawler visited; 0% means crawlers visited and none
could be verified. Those are opposite findings. See
How reliable is it.
Bot filtering does not apply here
Every card on this panel ignores the "Hide bots" toggle, on purpose. Turning bots off would empty the one feature you came for.
The AI panel's own filter is crawler type, so you can narrow the whole dashboard to, say, AI training traffic only.
Not the same as the AI channel
Your Traffic sources → Channels card has an entry called AI. That is a different measurement:
The AI channel counts people who clicked a link inside a ChatGPT, Perplexity or Claude answer and landed on your site.
This panel counts crawlers fetching your pages. No human involved.
Supply and demand. Read them together and you get "the assistants are reading me, and here is whether that turns into visits" — which neither answers alone. See Channels.