Bot intelligence
API
2026
Any client can claim to be GPTBot
A user agent is just a string, and faking a crawler takes one line of code. Telling real bots from impostors means maintaining IP range lists, running reverse DNS lookups, and parsing headers yourself, for thousands of bots and AI agents that change constantly. Most teams either guess or don't bother, and the tools that do it well lock you into routing all your traffic through their network.
One verdict, off the critical path
Send request details to a single endpoint as a side call, so your response never waits on us. The service responds with one verdict covering identity, delegation, verification, headers, network, and automation, plus a risk score and a recommended action. Request the full verdict or only the fields you need. It works from any stack over REST, with a JavaScript SDK for Node and Cloudflare Workers that adds caching and fails open, so your site keeps serving visitors no matter what.
Evidence, not just a flag
CDN bot management makes you route traffic through the vendor. Agent directories give you a verified or spoofed flag with no evidence behind it. IP intelligence APIs know the network but not who's on it or why. Bot Insights combines identity, purpose, verification, and network evidence in one API that works alongside the CDN, framework, and language you already use.
Unverified is not the same as spoofed
Not every bot publishes IP ranges, reverse DNS, or signed identity, so a missing check can't be treated as proof of a lie. Labeling a legitimate crawler as spoofed gets it blocked and costs the site its search or AI visibility. Every verdict separates what was proven from what couldn't be checked, so bots that can't be verified are marked unverified instead of spoofed, and the risk score reflects the actual evidence.