Can AI find and cite your site?
Enter your domain. We check whether AI crawlers can reach it, whether you have an llms.txt, and if your content is legible to agents that don't run JavaScript.
The dashboard is not the point. This is.
Your agent reasons over traffic and hands you the answer, already summarized. No tab switching, no scanning charts.
Everything your agent needs.
“How is my site doing right now?”
One compact overview: visitors, pageviews, top movers, and what changed for the period. The first thing your agent reads.
get_summary
“Show me X, broken down by Y.”
Any single metric, filtered and grouped: signups by referrer, pageviews by country, bounce on one path.
query_metric
“What changed after I shipped?”
This week vs last, pre-deploy vs post. Deltas come back with significance hints, so the agent knows what is noise.
compare_periods
“Why did the numbers move?”
Spikes and drops surfaced with a probable cause attached: a vanished referrer, a broken page, a crawler wave.
detect_anomalies
“Is my traffic real, or bots?”
Human visitors and crawler hits split cleanly, plus which AI products are sending you actual people.
get_ai_traffic
“What should I fix to get cited?”
Ranked, evidence-backed actions: pages AI crawlers cannot fetch, cited pages that bounce, winners worth amplifying. Shaped for an agent to act on.
get_recommendations
“What is working best?”
Top pages, referrers, countries, devices, or events, ranked with share of total and week-over-week movement built in.
list_top
A crawler is not a customer.
Windrose splits AI crawlers from real humans who arrived from an AI answer, then groups crawler hits by inferred purpose: training, indexing, retrieval, active-agent, or unknown. The JavaScript beacon alone undercounts crawlers that do not run JS, so one small server-side middleware captures recognized AI crawlers at the edge.
Bots fetching your pages for training, indexing, retrieval, or a user-directed action. Useful to watch, never a conversion.
Real people who clicked through from an AI answer. This is the traffic that converts.
The JavaScript beacon only sees crawlers that run JS, so non-JS bots like GPTBot and ClaudeBot slip past it. Drop in the server-side middleware to capture recognized AI crawlers. Integrations that observe the final response also record its HTTP status, so you can see a bot hit a 404 after a deploy, fix the page, and watch the next crawl come back 200. See the setup.
No tracking cookies. No raw IP in analytics.
Windrose minimizes visitor data by design. The tracker builds no cross-site profile, raw IP addresses are discarded after country and daily visitor derivation instead of entering the analytics dataset, and the resulting identifier rotates every day.
Priced for people who ship, not enterprises.
Free while we are in beta
These tiers are a preview of launch pricing. Until then, everything is free.
Join the beta