Anthropic released a beta feature on October 9, 2026, that allows Claude Managed Agents to write programs coordinating multiple background agents. This update enables agents to define phased work, execute parallel tasks, and repeat rounds based on intermediate results. The feature builds on the multiagent orchestration beta launched in May 2026, giving agents more control over complex workflows. Developers must configure the agent to start runs and monitor execution through specific event streams. Operational limits include a 24-hour run lifetime and a maximum of 10 open runs per session.
Read the full story at ai.info.
Photo: One Idea LLC (CC0) via Openverse
