AI BriefingAnthropic launches self-hosted Claude Managed Agents in private beta
·2026.11.13 11:35
Key point
The feature allows running agent harnesses and sandboxes in customer environments while routing model requests to Anthropic, Bedrock, Google, Microsoft, or custom gateways.
Details
- Anthropic announced self-hosted Claude Managed Agents, allowing the harness, sandboxes, credentials, and session history to run in customer cloud or on-prem environments.
- Model requests can be routed to Anthropic directly, Amazon Bedrock, Google's Agent Platform, Microsoft Foundry, or a customer's own gateway.
- Security architecture keeps the agent loop and session history outside the sandbox, uses a proxy to attach credentials so secrets never enter the sandbox, and applies host allowlists per environment.
- Access is currently limited to customers working with an Anthropic account team, and zero data retention applies only if the organization already has that agreement with Anthropic.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.