Anthropic has launched a public beta for Claude Managed Agents, a cloud infrastructure service that simplifies creating and running autonomous AI agents. The service handles backend tasks such as sandboxes, session states, access control, and orchestration.

Key features include support for long autonomous sessions with context preservation, multi-agent coordination for parallel tasks, and an iterative self-evaluation algorithm to refine results. Deployment is available via the Claude web console, CLI, or Claude Code.

Pricing combines standard token fees with an additional $0.08 per hour of active runtime.

For more details, visit claude.com.