The logs your stack already writes, read the moment they are written. No pixel, no snippet, no sampling. The cleanest first-party account of every crawler that touches your pages.
Every crawler that hits you, named and identity-checked against the addresses it claims to own. Impersonators are caught in the same stream.
| Crawler | Hits | Identity |
|---|---|---|
| GPTBot | 1,284 | Verified |
| ClaudeBot | 611 | Verified |
| PerplexityBot | 330 | Verified |
| Googlebot (claimed) | 417 | Fake |
All four run from the same real-time stream, so there is nothing extra to set up per use case. Connect once and every signal arrives together: search crawlers, AI crawlers, errors, latency and abuse, from the logs you already produce.
Real requests on a demo property, updating as they land. Click through and use it — no account, no sales call. Test the live demo →
The same broken deploy, told two ways. Batch log tools collect overnight and report at nine. We classify every request as it arrives and alert inside five minutes.
76% from one Singapore address using twenty user-agents. Your canonicalisation is fine. Rate-limit it at the CDN.
sqlmap and /.env probes from three addresses. One returned a 200, so that resource may really exist.
GPTBot up 4.8×, 998 hits across /blog, /products and /pricing.
Each alert runs a reachability probe first, so we never tell you the site is down when it is not.
The same crawl data legacy tools sell at $75k a year. Unlimited users on every plan.
Free tier, no card, five minutes to your first request.