Wisp

Serverless for AI agents.

Agents make millions of tiny tool calls per day. Today's serverless platforms add ~1.5 seconds of cold-start overhead per call — that's 75 seconds of pure friction in a 50-call agent task.

We make those calls fly: ~1ms cold start for 80% of agent calls via WASM. Native fallback for the rest, when you need the full Python ecosystem.

Built for the agent era.