Anthropic Adds Dynamic Workflows to Claude Managed Agents, Enabling 1,000 Parallel AI Agents

How Dynamic Workflows Work

Anthropic is adding dynamic workflows to Claude Managed Agents, bringing multi-agent orchestration to its platform. The managed agent infrastructure has existed for some time, but dynamic workflows are new. In this setup, a lead agent creates a plan, distributes tasks to sub-agents, and merges the results once they are done. Up to 1,000 agents can run in parallel per execution.

Anthropic's Internal Bug-Hunt Test

Anthropic evaluated the feature by hiding 70 bugs in a 116,000-line codebase. A single agent caught between 14 and 27 bugs per run, while the dynamic workflow consistently found 66. The company acknowledges that whether these gains hold across different task types remains to be seen, so users are advised to validate the approach with their own workloads.

Cost Concerns and How to Enable

Whether the approach is cost-effective is debatable. A senior OpenAI engineer recently described agent swarms as a massive waste of tokens. Anthropic concedes that these workflows can consume a lot of tokens and recommends starting small. To activate dynamic workflows, users select the "multiagent_20261001" agent type. Setup is available through Anthropic's documentation or by running "/claude-api managed-agents-onboard" in Claude Code.

Sources: The Decoder