Anthropic launches Claude Managed Agents to accelerate AI agent development & deployment

Anthropic has introduced Claude Managed Agents, a new offering that provides composable APIs for deploying cloud-hosted agents at scale. This release allows developers to focus on defining agent tasks and guardrails while Anthropic handles infrastructure, security, and state management, reducing deployment time from months to days.

Agents deployed with Managed Agents run on production infrastructure and benefit from secure sandboxing, authentication, and automatic state persistence across session interruptions. While these agents are active autonomously for extended periods, progress and outputs are maintained even during disconnections. Trusted governance features include scoped permissions, identity management, and execution tracing, supporting enterprises that require robust oversight and auditability.

Following these foundational improvements, Managed Agents enables multi-agent orchestration. This feature, currently in research preview, allows multiple agents to coordinate and parallelize complex workflows. Built to leverage Claude models, the platform supports iterative agent execution by enabling agents to self-evaluate against defined success criteria or operate in a traditional prompt-and-response mode as required.

Session tracing, integration analytics, and troubleshooting tools are integrated within the Claude Console, making it possible for developers to inspect tool calls and agent decisions in detail. Managed Agents is available now in public beta on the Claude platform.