Meta AI
Meta
41,180
requests · 7d
CFI.co · AI Visibility
The world's AI is reading CFI.co
Between 1 July and 30 July 2026, AI engines made 727,934 requests to CFI.co content — crawling it, indexing it, and fetching it to answer questions.
In the last 7 days: 171,764. The per-engine breakdown below covers that 7-day window, because Cloudflare's per-agent data only reaches back eight days; the 30-day total is accumulated from our own daily record.
Last updated · rebuilt from Cloudflare every 20 minutes
Evidence: first-party Cloudflare edge request logs. Strength: strong (direct access).
What this measures — and what it cannot
These are requests that identified AI systems made to CFI.co's own servers, logged at our edge. They are a floor, not a total. They do not include AI systems reading our content from our public GitHub archives, which we have no way to measure; answers generated from material already absorbed into a model during training, which require no fetch at all; answers drawn through Google's search index rather than by a branded Gemini fetcher; or any system that does not identify itself honestly. Every one of those gaps runs the same way — AI use we cannot see, rather than use we have overstated. We cannot tell you how large they are, and we would rather say so than estimate.
The largest of those gaps exists because of choices we made deliberately. We mirrored our archives to GitHub so machines could read us cleanly, and we opened robots.txt rather than gating access. Both improved machine access and destroyed our visibility into it. We cannot maximise both, and we chose the reader over the metric.
The 30-day total is accumulated from our own daily measurements. Readings before 20 July 2026 were rebuilt from our server's own daily log after we found and corrected a fault in how we stored them.
AI engines crawling CFI.co
Each engine identified by its published crawler user-agent. Training = building models · AI Search = answering queries · User fetch = a user asked the AI about this page.
Doubao
ByteDance
41,108
requests · 7d
Amazon
Amazon
38,772
requests · 7d
ChatGPT
OpenAI
35,690
requests · 7d
Claude
Anthropic
10,749
requests · 7d
You.com
You.com
1,854
requests · 7d
Perplexity
Perplexity
1,549
requests · 7d
Gemini
494
requests · 7d
Why Gemini's figure looks modest: Gemini grounds its answers on Google's Search index, and Googlebot crawls CFI.co 48,818 times a week to build that index — so CFI.co is firmly in the index Gemini draws from, even though Gemini's own branded fetchers are light.
DuckDuckGo AI
DuckDuckGo
184
requests · 7d
Common Crawl
Common Crawl
178
requests · 7d
Apple Intelligence
Apple
4
requests · 7d
Cohere
Cohere
2
requests · 7d
Why Gemini's figure looks modest: Gemini grounds its answers on Google's Search index, and Googlebot crawls CFI.co 48,818 times a week to build that index — so CFI.co is firmly in the index Gemini draws from, even though Gemini's own branded fetchers are light.
AI crawl activity over time
One bar per complete UTC day, measured exactly as the figures above. They add up to 727,934 — the 30-day total.
AI assistants fetching CFI.co while answering
When people ask ChatGPT, Perplexity and others, these are the times the assistant fetched CFI.co in real time to answer them — live query activity, distinct from background training crawls. What happens next is outside our view: we can see that the content was retrieved, not whether the answer went on to use or cite it. (AI tools strip referrer headers, so we measure this from their own published fetch agents — a more reliable signal.)
- ChatGPT18,518 fetches
- Claude2,060 fetches
- You.com1,854 fetches
- Perplexity1,549 fetches
- Gemini494 fetches
- DuckDuckGo AI184 fetches
Open to AI by design
CFI.co welcomes every major AI crawler: we block no AI crawler. Many publishers block them; CFI.co is openly part of the AI knowledge layer by choice.
A small number of paths — order forms and pages that are not yet live — are withheld in robots.txt from everyone, human and machine alike. No AI crawler is given narrower access to CFI.co than any other visitor.
How this is measured (proof of source)
Live data from the Cloudflare GraphQL Analytics API for zone cfi.co. The first query returns every user-agent seen at our edge in the window; we then classify those user-agents against the published AI-crawler list and sum the AI ones. General search crawlers (Googlebot, Bingbot) are classified separately and are NOT included in any AI figure on this page.
The engine figures and the 7-day total come from this query:
query{viewer{zones(filter:{zoneTag:"5c25ec4a465cb7c5b0f1131cfd016ce9"}){httpRequestsAdaptiveGroups(limit:10000,filter:{datetime_geq:"2026-07-24T18:20:01Z",datetime_leq:"2026-07-31T18:20:01Z"},orderBy:[count_DESC]){count dimensions{userAgent}}}}}This one runs once per completed day; the 30-day total is the sum of thirty of them:
query{viewer{zones(filter:{zoneTag:"5c25ec4a465cb7c5b0f1131cfd016ce9"}){httpRequestsAdaptiveGroups(limit:10000,filter:{datetime_geq:"2026-07-30T00:00:00Z",datetime_lt:"2026-07-31T00:00:00Z"},orderBy:[count_DESC]){count dimensions{userAgent}}}}}