What runtime API metering means
UsageFlow is AI runtime billing infrastructure: it meters API and AI usage inside your app, enforces spend and rate policy before costly work executes, and can sync usage to Stripe. See AI usage metering and AI payment infrastructure.
Why gateway-only control falls short
Gateways are useful at the HTTP edge. They are a weak sole answer for modern APIs and AI products:
- Edge ≠ runtime — retries, tool calls, and function work happen after the gateway already said yes.
- Static configs lag product — every new endpoint or plan becomes another manual change.
- DIY glue — teams rebuild metering, throttling, and billing export in-house and maintain it forever.
Engineers refuse another metering subsystem. FinOps needs enforceable spend. Founders need usage that can land cleanly in Stripe. Runtime metering is the shared answer.
How UsageFlow tracks, meters, and throttles
- Install a framework agent before your routes
- Set monitoringPaths and whitelistEndpoints in Console
- Apply policies that meter and enforce before execution
- Optionally configure a Stripe API key to fire meter events
Related: meter your first request · meter AI agents · sync to Stripe.
Meter · enforce · bill
- Meter — capture API and AI usage in runtime
- Enforce — throttle or block when policy says stop
- Bill — ship billing-ready events to Stripe Meters
FAQ
How is runtime metering different from an API gateway?
Gateways work at the HTTP layer. UsageFlow agents sit in your app runtime, so you can meter and enforce work that gateways never see — including AI fanout after the request is accepted.
Do I still need a gateway?
Often yes for edge concerns. UsageFlow does not replace every gateway job; it meters and enforces in runtime where spend and agent behavior actually happen.
Where do route lists live?
In Console Configuration — monitoringPaths and whitelistEndpoints — not hard-coded as the source of truth in your install snippet.