Anthropic rolls out Claude Managed Agents

Anthropic has announced Claude Managed Agents, a suite of composable APIs for building and hosting cloud-hosted agents. The intent is to give any company building on the Claude Platform the full production stack for shipping AI agents at scale.

Launched April 8 in a public beta on the Claude Platform, Claude Managed Agents is purpose-built for Claude, enabling better agent outcomes with less effort by providing sandboxed code execution, checkpointing, credential management, scoped permissions, and end-to-end tracing, according to Anthropic. Users define an agent’s tasks, tools, and guardrails and deploy it on Anthropic infrastructure. A built-in orchestration harness decides when to call tools, how to manage context, and how to recover from errors, the company said.

Claude Managed Agents includes the following:

• Production-grade agents with secure sandboxing, authentication, and tool execution handled for the user.
• Long-running sessions that operate autonomously, with progress and outputs that persist through disconnections.
• Multi-agent coordination that allows agents to spin up and direct other agents to parallelize complex work.
• Trusted governance that gives agents access to real systems with scoped permissions, identity management, and execution tracing built in.

Notion, Rakuten, and Sentry are already building on Claude Managed Agents, Anthropic said. In internal testing around structured file generation, Claude Managed Agents improved outcome task success by as much as 10 points over a standard prompting loop, with the largest gains on the hardest problems, Anthropic said.