Claude Managed Agents on Vercel's Agentic Infrastructure
summary
Demonstrates running Claude via Anthropic’s cloud-managed agent control plane while executing code inside Vercel sandboxes so network policy and data never leave customer infrastructure. Claims: the control plane handles sessions, logging and model access while execution runs in Vercel sandboxes; each session gets one sandbox per session that starts in seconds. Tooling uses an event stream API, MCP tunnels for private services, and outputs land in a Vercel workspace; the hybrid setup is in public beta and documented.
AI-written summary · about
chapters (7)
- 0:00Introduction
- 1:09Cloud managed agents
- 2:31Constraints and hybrid
- 3:56Splitting with Vessel
- 7:07Why Vessel sandboxes
- 9:34Building with managed agents
- 12:43Summary








