LangChain revamps async subagent support in Deep Agents and Managed Deep Agents

LangChain announced on Nov. 5 that it had revamped support for async subagents in Deep Agents and Managed Deep Agents. The company says the subagents can make progress on background tasks while keeping the orchestrator agent responsive to users. The post does not provide performance measurements for that claim.

In a separate post that day, hwchase17 said most agents being built inside LangChain use async subagents “in a meaningful way” and that they are now available in Deep Agents. Neither post explains how to configure the support.