AI BriefingLangChain Adds Async Subagents to Deep Agents for Parallel Background Work
·2026.11.06 05:00
Key point
The feature is available in Python and JavaScript, with task notifications currently in beta.
Details
- LangChain introduces async subagents to Deep Agents, allowing an orchestrator to remain responsive to users while background workers handle long-running tasks in parallel.
- This architecture resolves the blocking issue in standard ReAct loops, where each tool call halts progress; it aligns with patterns used by Claude Code, OpenAI's dots, and DeepSeek Harness.
- The orchestrator manages background work via five tools: start_async_task, check_async_task, update_async_task, cancel_async_task, and list_async_tasks.
- Steering running subagents uses an update_strategy field with "interrupt" (default, stops and restarts) or "enqueue" (waits for current run to finish) modes.
- Notifications are in beta; the system handles results based on orchestrator state (busy, idle, or paused) and supports approval workflows on subagent threads.
- Available in Deep Agents for Python and JavaScript, and Managed Deep Agents, where enabling async requires setting the "async" field to True.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.
Related
OpenCode 2.0 Released with Centralized Service, Code Mode, and Background TasksMicrosoft Engineers Detail Copilot Cowork's 4.5-Month Path to GA and Python-to-Rust Migration