AI report

See what AI reads, and who it sends you.

Crawlers read your pages for training, search or someone’s question. Referrals are the people who click through from an AI answer. You see both, side by side.

Start 30-day trial Set up the edge collector

Shows crawlers and visitors from

ChatGPT
Claude
Perplexity
Gemini
Copilot
Meta AI
Mistral
DeepSeek
Grok

Hover a name for the bots we match. Trademarks belong to their owners. Not affiliated.

AI crawler requests8,494+18%
AI crawlers seen6
Visitors from AI assistants1,749+31%
Search engine crawler requests11,820for comparison

AI crawler requests over timeTrainingSearchUser

AI crawlersRequests per bot, counted by the edge collector. Not people.

GPTBot OpenAITraining3,184
ClaudeBot AnthropicTraining2,207
OAI-SearchBot OpenAISearch1,342
PerplexityBot PerplexitySearch961
ChatGPT-User OpenAIUser request412
Bytespider ByteDanceTraining388

AI assistantsPeople who clicked through from an AI answer

chatgpt.com70%1,218
perplexity.ai20%344
claude.ai11%187

Also counts utm_source=chatgpt.com and similar tags.

Which pages AI crawlers readPer pagePer page and bot

/blog/ideal-vs-creditcard/1,904
/products/1,602
/pricing/1,034
/returns/498
/blog/976

Example data for an invented shop. The layout matches the AI report in the app.

Two different things

Bots aren’t visitors. We count them apart.

Crawlers

Requests, not people

AI crawlers don’t run JavaScript, so a script never sees them. The edge collector counts them on the server, per page and per bot.

Needs the edge collector

Referrals

People from AI answers

Visitors from chatgpt.com, perplexity.ai, claude.ai and others form one channel. You see what they read and which goals they reach.

Works with the script alone

What you learn

Questions it answers.

01
Is AI training on my best content?
Training crawlers per page show which articles and product pages are being collected for training.
02
Can AI search find my pricing page?
Search crawlers like OAI-SearchBot and Claude-SearchBot show what assistants can cite. If one never visits, that page won’t be quoted.
03
What are people asking AI about us?
A user-request fetch happens when someone asks an assistant something that needs your page. Those pages show what people ask about.
04
Which assistant sends real visitors?
ChatGPT, Perplexity, Claude, Gemini and Copilot referrals are one channel, with their landing pages and goals.
05
Is my robots.txt doing what I think?
See whether a bot you blocked still shows up, and which ones you allowed actually come.
06
Is AI replacing search for my site?
AI crawler requests sit next to Googlebot and Bingbot, so you can compare the two over time.

What we recognise

26 AI crawlers, 17 assistants.

This list comes from our source code, which has tests. If a crawler is missing, it shows up as an unknown bot, not as a person.

TrainingReads pages to train models.

GPTBotOpenAI
ClaudeBotAnthropic
anthropic-aiAnthropic · legacy
Google-ExtendedGoogle · robots.txt only
GoogleOtherGoogle
Applebot-ExtendedApple · robots.txt only
BytespiderByteDance
CCBotCommon Crawl
Meta-ExternalAgentMeta
cohere-training-data-crawlerCohere
AI2BotAllen Institute for AI
DiffbotDiffbot

AI searchIndexes pages so an assistant can cite them.

OAI-SearchBotOpenAI
Claude-SearchBotAnthropic
PerplexityBotPerplexity
AmazonbotAmazon
YouBotYou.com

User requestFetches a page because a person asked about it.

ChatGPT-UserOpenAI
Claude-UserAnthropic
Claude-WebAnthropic · legacy
Perplexity-UserPerplexity
Google-CloudVertexBotGoogle
Meta-ExternalFetcherMeta
DuckAssistBotDuckDuckGo
MistralAI-UserMistral AI
cohere-aiCohere

For comparison we also count Googlebot, Bingbot, Applebot, DuckDuckBot, YandexBot, Baiduspider. Crawlers are matched by user agent; we don’t check IP ranges, so a bot pretending to be GPTBot counts as GPTBot.

Limits

What it can’t see.

robots.txt tokens: Google-Extended and Applebot-Extended control how your content is used, but Google and Apple crawl as Googlebot and Applebot. There is nothing separate to count.
Answers without a click: If an assistant quotes you and nobody clicks, there’s no referral. A user-request fetch is often the only trace.
Without the edge collector: With only the script you get AI referrals but no crawlers, because crawlers don’t run JavaScript.

GA4, Plausible and Fathom also have an AI channel for referrals (Analytics Help: Default channel group; Plausible: Changelog; Fathom changelog: AI tab (Oct 2026); checked Nov 2026). What Wholetally adds is the crawler side. See the comparison.

Count the visitors you’ve been missing.

30-day free trial, no card needed. Create an account with your work email and add your first site.

Start 30-day trial