Anthropic updates Claude Managed Agents with three new features
9to5Mac • • technology
Key Points:
- Anthropic has enhanced Claude Managed Agents with three new features: dreaming, outcomes, and multiagent orchestration, aimed at improving AI agent capabilities and ease of deployment.
- Dreaming is a research preview feature that reviews past sessions to identify patterns and refine agent memory, allowing agents to self-improve over time either automatically or with user review.
- Outcomes enable users to define success criteria for agents through a rubric, with a separate grader evaluating agent performance and providing feedback for iterative improvement, along with webhook notifications upon task completion.
- Multiagent orchestration allows a lead agent to break tasks into parts and delegate them to specialized subagents working in parallel, sharing context and contributing to the overall workflow, enhancing efficiency in complex tasks.
- Companies like Netflix are already utilizing Claude Managed Agents with multiagent orchestration to support their platform teams, demonstrating practical adoption of these new features.
