By late 2025 there was one AI bot visit for every 31 human visits (TollBit). But lumping them all into "block or allow" ignores the only question that matters for your business: which ones are actually valuable?
Most platforms treat AI traffic as a binary: threats to block, or crawlers to allow. That framing misses the entire story.
AI bots hitting WooCommerce cart pages 3.75 million times in a single day (Kinsta) is a different problem than ChatGPT fetching your product page because a real customer asked about it. Blocking both loses you the valuable traffic. Allowing both burns your server budget.
The question isn't "should I allow bots?" — it's "which bots, on which parts of my site, and what are they worth to me?"
Answering that requires intelligence, not a firewall.
CortIQ classifies every AI visit so you can act on the right signal — not just know that "bots visited."
Crawling your content to train AI models. Generates server load, zero referral traffic.
Measure the cost. Decide if you want to allow or restrict.
An AI assistant opening your page because a real user asked it to. The user has intent — the AI is their browser.
Track their journey. Attribute conversions. Optimize for them.
Indexing your content for AI-powered search results. Drives indirect discovery in ChatGPT, Perplexity, Gemini.
Monitor access. Optimize for citability.
Bot blockers and CortIQ solve different problems. You likely need both — but for different reasons.
| Capability | Bot blockers | CortIQ |
|---|---|---|
| Detect AI bot visits | ||
| Classify by bot type (training / agentic / citation) | ||
| Track agentic fetches separately from crawlers | ||
| Measure conversion from AI-referred users | ||
| AI visibility & citability scoring | ||
| Citation request tracking | ||
| Block bots at the edge |
Not just bot detection — bot intelligence.
Every AI request is classified at ingest as training, agentic or citation — the same rules for the JS tag and the Cloudflare Worker.
When an AI assistant opens a page for a user, it identifies itself with a -User token. CortIQ tracks those as real-intent traffic.
See how far AI agents get through your site — from landing page to conversion.
Dedicated dashboards for AI-driven traffic — split cleanly from human visitor data.
Training and citation crawlers rarely execute JavaScript. A Cloudflare Worker reports them and CortIQ classifies them server-side with the same registry.
Every AI recommendation shows its work — see the exact tables, row counts, and model behind each insight. No black box.
The 27 named bots CortIQ classifies today, straight from the ingest registry
| Agent | Vendor | Category |
|---|---|---|
| ChatGPT-User | OpenAI | Agentic Fetches |
| ChatGPT Atlas | OpenAI | Agentic Fetches |
| OAI-SearchBot | OpenAI | Citation & Search Crawlers |
| GPTBot | OpenAI | Training Crawlers |
| Claude-User | Anthropic | Agentic Fetches |
| Claude-SearchBot | Anthropic | Citation & Search Crawlers |
| ClaudeBot | Anthropic | Training Crawlers |
| Perplexity-User | Perplexity | Agentic Fetches |
| PerplexityBot | Perplexity | Citation & Search Crawlers |
| Google-Extended | Training Crawlers | |
| GoogleOther | Citation & Search Crawlers | |
| Googlebot | Citation & Search Crawlers | |
| Bingbot | Microsoft | Citation & Search Crawlers |
| Meta-ExternalAgent | Meta | Training Crawlers |
| FacebookBot | Meta | Citation & Search Crawlers |
| Applebot-Extended | Apple | Training Crawlers |
| Applebot | Apple | Citation & Search Crawlers |
| YouBot | You.com | Citation & Search Crawlers |
| DuckAssistBot | DuckDuckGo | Citation & Search Crawlers |
| Grok | xAI | Agentic Fetches |
| Amazonbot | Amazon | Citation & Search Crawlers |
| Bytespider | ByteDance | Training Crawlers |
| CCBot | Common Crawl | Training Crawlers |
| Diffbot | Diffbot | Training Crawlers |
| cohere-ai | Cohere | Training Crawlers |
| DeepSeek | DeepSeek | Training Crawlers |
| MistralAI | Mistral AI | Training Crawlers |
Free during the beta. Add one script tag or the WordPress plugin.