Anthropic has introduced Claude Managed Agents, a new cloud service aimed at reducing the complexity of building production-ready AI agents.
The key promise is speed. What previously took months of engineering effort can now be done in weeks.
Building an agent has never just been about the model. Teams also had to set up containers, infrastructure, observability, security rules, and state management. Most of the effort lived outside the agent itself.
Claude Managed Agents consolidates that work into a set of APIs.
Developers define:
the task
the tools the agent can use
the security constraints
From there, the system handles the rest.
Each agent runs in an isolated container with preconfigured environments. State management, including handling external data and sensitive credentials, is managed by the platform.
Tool orchestration is built in, allowing the system to decide which tools to use for a given task.
The service also introduces reliability features. Agents can recover from interruptions and resume work without starting over.
Two capabilities are currently in research preview:
agents that can spawn other agents for complex workflows
automatic refinement of responses, which Anthropic reports improves task success rates in internal testing
Pricing is usage-based, combining model costs with a fixed runtime fee.
Early adopters include companies like Notion Inc., Rakuten Group Inc., and Asana Inc., some of which have already integrated these agents into their products.
