Anthropic launched Claude Managed Agents last month, significantly simplifying the work required to create and deploy cloud-hosted AI agents. This week, Claude managed agents become more efficient with three new features.
Anthropic Launches the Dream, Results and Multi-Agent Orchestration for Claude Managed Agents
The first new feature is called dreaming, which Anthropic classifies as a search preview. Anthropic says that dreaming expands Claude’s memory abilities “by examining past sessions to find patterns and help agents improve.”
Dreaming is a planned process that examines your agents’ sessions and memories, extracts patterns, and organizes the memories so that your agents improve over time. You decide how much control you want: dreaming can update the memory automatically or you can review changes before they appear.
Anthropic describes how memory and dreaming work together to improve Claude Managed Agents:
Together, memory and dreaming form a robust memory system for self-improving agents. Memory allows each agent to capture what it learns how it works. Dreaming refines this memory between sessionsby drawing on learnings shared between agents and keeping them up to date.
Meanwhile, Outcomes is a new feature in Claude Managed Agents that allows you to explain what defines a successful outcome for the agent to accomplish.
With the results, you write a rubric describing what success looks like and the agent works on it. A separate evaluator evaluates the result against your criteria in its own pop-up, so it isn’t influenced by the agent’s reasoning. When something goes wrong, the evaluator identifies what needs to change and the agent makes another pass. (…)
You can also now set a result, let the agent run, and be notified by a webhook when the operation is complete.
Finally, there’s Multi-Agent Orchestration, a new Claude Managed Agent tool that “allows a master agent to divide work into multiple chunks and delegate each to a specialist with their own template, prompts and tools.”
For example, a master agent can investigate while sub-agents fan out into deployment history, error logs, metrics, and support tickets.
These specialists work in parallel on a shared file system and contribute to the overall context of the lead agent. The master agent can communicate with other agents during the workflow because events are persistent and each agent remembers what it did.
Anthropic describes how agents managed by Claude are already used by companies like Netflix, which has already deployed multi-agent orchestration for its platform team.
You can learn more about Claude Managed Agents and the three new features Anthropic launched this week.
FTC: We use automatic, revenue-generating affiliate links. More.