Dynamic Subagents: How to Run Parallel Agents Reliably in Deep Agents
LangChain launched dynamic subagents in Deep Agents, letting agents spawn and coordinate parallel subagents by writing code.
“So the orchestration effectively moves out of the agent's head and into code.”
LangChain released dynamic subagents for Deep Agents, a feature that lets an agent write code to programmatically spawn and coordinate many parallel subagents via a code interpreter middleware and a 'task' global. This matters because it moves orchestration logic out of the agent's reasoning and into deterministic code, improving reliability and completeness for large-scale, repetitive multi-step tasks.