See which AI assistants and crawlers read your site.

• GPTBot

  OpenAI's training/index crawler. Distinguished from human ChatGPT referrals.

• ClaudeBot

  Anthropic's crawler. Honored separately from anthropic.com human visits.

• PerplexityBot

  Perplexity's retrieval crawler that powers cited answers.

• Google-Extended

  Google's AI-training opt-out signal-aware crawler.

• Applebot

  Apple's crawler for Siri/Spotlight and increasingly AI surfaces.

• Bingbot

  Microsoft's crawler — also a frequent backbone for Bing-powered AI answers.

• Meta-ExternalAgent

  Meta's external-data crawler used for model training.

• CCBot

  Common Crawl bot, the corpus that underpins many open AI projects.

The signature catalog lives in @webmasterid/ai-visibility and grows over time. New AI surfaces are added as they appear with stable user-agent strings.

• Human ChatGPT/Claude/Perplexity referrals

  When an AI assistant cites or links your site, that referral arrives as a real human visit with a recognisable referrer. WebmasterID classifies it as ai_referral so you can size demand from each surface.

• Crawler observability

  AI training and retrieval crawlers visit on their own schedule. Knowing which models are reading what lets you set robots/sitemap policy with real data rather than guesswork.

• Programmatic SEO health

  Large content networks need to know which pages are getting indexed by which crawlers. WebmasterID makes that a queryable surface.

• Publisher economics signal

  If your content is being summarised inside AI surfaces without a referral, that's a signal too. Bot visits without matching human referrals tell a story.

Every event arriving at the ingest API is classified at the edge. Requests with a recognised AI/search-bot user-agent get persisted to the bot-visit path; everything else goes to the events path. The dashboard surfaces both — on its own page (AI Visibility), in tracking-health summaries, and in the per-page breakdown.

Human visits that arrive with an AI-assistant referrer (e.g., a ChatGPT citation click, a Perplexity outbound) are kept on the human side but tagged as ai_referral so you can see which AI surfaces are sending real readers.