Anthropic Releases Self-Hosted Claude Managed Agents

The private beta lets customers run agent infrastructure and session history within their own cloud or on-premise environments.

Updated on Nov. 12, 2026 in Artificial Intelligence

Bold vector editorial illustration of a row of slate-blue enclosures with a coral cable bundle running out of an open floor panel, representing self-hosted AI agent infrastructure.
Anthropic has released a self-hosted option for Claude Managed Agents in private beta, letting customers run agent infrastructure in their own cloud or on-premise environments. AI Illustration. Upload story photo >

Anthropic has released a self-hosted version of Claude Managed Agents, allowing customers to run the agent harness, sandboxes, credentials, and session history in their own cloud or on-premise. The self-hosted option opened as a private beta on November 12, and customers request access through their Anthropic account team.

Why it matters

Until now, customers could host only the sandbox themselves, so the expanded option gives organizations more control over where agent infrastructure and data reside. Developers build agents with the same core APIs as the hosted service.

Session history is stored in the customer's own PostgreSQL database, and the customer operates the Managed Agents server and the workers where agents run shell and file tools. Claude itself still runs outside the customer's network, and the extraction does not say how latency or pricing differ from the hosted service.

The players

Anthropic

Anthropic is an AI safety company that develops the Claude family of large language models and offers them through its own API and cloud platform partners.

The details

In the self-hosted setup, customers operate the Managed Agents server, its PostgreSQL database, and the workers that run agent tools, while model requests can route to Anthropic directly, Amazon Bedrock, Google's Agent Platform, Microsoft Foundry, or a customer's own gateway. Anthropic built the system around four security principles: for MCP tools and supported API calls, a proxy outside the sandbox attaches credentials to requests, and each environment can be locked to an allowed host list that blocks all other destinations.

Timeline

  1. November 12, 2026: Self-hosted option opens as a private beta.

The Tech Race

The self-hosted release extends the hosted Claude Managed Agents service by moving the agent harness, credentials and session history into customer-controlled infrastructure. It positions Anthropic to compete on deployment flexibility as enterprises weigh where autonomous agents should run.

Organizations that need agent data to stay inside their own infrastructure can request private beta access through an Anthropic account team. Developers can reuse the same core APIs as the hosted service, though the self-hosted option requires operating the server, database and worker environments.

The takeaway

Teams evaluating autonomous agents should weigh where session data and credentials live as carefully as which model they use. Self-hosting shifts operational responsibility to the customer, so planning for databases and worker infrastructure matters early.

Further reading

Explore more Artificial Intelligence coverage for context on how agent platforms are evolving.