CrowdStrike's Falcon AI Gateway now generally available

CrowdStrike built Falcon AI Gateway for its own internal use first, drawing on its AI Hub gateway that has served tens of thousands of employees globally over the past few years. That experience shaped the product's design: the company learned to keep the request path short, since every extra hop adds another failure point and more time. It also found that most teams would rather not run a gateway themselves, so it is now offering the one it built for its own needs to Falcon Guardian customers.

The gateway complements existing discovery tools by making the approved path the only path, but it does not automatically discover or redirect AI traffic. Instead, teams must block direct connections to model providers using network controls and deliberately route workloads through the gateway. Once traffic is routed, administrators can switch between connected model providers without reconfiguring every workload, and apply rate limits and budgets scoped by application, user, team, model, or provider.

Falcon AI Gateway and the Falcon sensor share one Falcon Guardian policy model, but each enforces where the policy sits. The sensor acts on the endpoint, including local processes, files, and commands an agent runs, while the gateway acts on routed traffic and cannot stop local processes. This division means the gateway extends enforcement beyond sensor coverage to cloud workloads and custom applications that call models directly through SDKs.