Managed Deep Agents: the fastest way to ship a production deep agent
The point
LangChain says Managed Deep Agents is in public beta on LangSmith Cloud. You build an agent as a folder of files, write its instructions and tools, pick the model, and deploy it with one command, with no servers to run.
Why it matters
For builders, LangSmith handles durable threads, memory, sandboxes, and scheduled runs.
Key details
- Only required file is agent.py; the system prompt lives in instructions.md
- Add memory.py for memory across threads and files under schedules/ for cron runs
- LangChain says Content Studio code shrank by about 80 percent after moving over
Original ~6 min → 25 sec here Source: LangChain Blog, Nov 2, 2026