Anthropic Unveils Claude Managed Agents to Cut AI Development Time from Months to Weeks

Anthropic has launched Claude Managed Agents, a new cloud service designed to help companies build and deploy AI agents faster, aiming to compress development timelines from months to just weeks.

"It pairs an agent harness tuned for performance with production infrastructure, so you can go from prototype to launch in days," Anthropic said.

Building production-ready AI agents typically requires extensive engineering work beyond the model itself, including setting up secure execution environments, infrastructure, and monitoring systems. Anthropic’s new offering automates much of this complexity, providing developers with a streamlined pathway via application programming interfaces (APIs).

With Claude Managed Agents, developers begin by describing the tasks they want automated and selecting the tools the agent should use. They can also define cybersecurity controls, such as whether certain actions require user approval.

The system then automatically provisions isolated containers for each agent, ensuring safe execution without risking core production systems.

The platform also simplifies state management, a critical component that governs how agents store and use data. This includes everything from integrating public web information into responses to securely handling sensitive data like login credentials.

Another key feature is automated tool orchestration. When an agent receives a prompt, the system determines which tools to deploy to generate the best response. It also includes error recovery capabilities, allowing agents to resume tasks after interruptions.

Anthropic is additionally testing advanced features in research preview, including multi-agent collaboration and automatic prompt refinement, which it says improved task success rates by up to 10 percentage points in internal evaluations.

Early adopters include Notion, Rakuten, and Asana, with several already integrating the service into their products.