LangChain opens Managed Deep Agents in public beta — one command from local test to hosted runtime

LangChain put Managed Deep Agents into public beta on 7 August 2026. A developer authors a Deep Agent in Python or TypeScript, tests it locally, and deploys it to a managed runtime with a single command, getting durable execution — a long-running agent can pause, retry and resume without losing work — plus sandboxes, persistence and memory, identity and multi-user auth, Slack and GitHub channels, scheduling and evals, without building that infrastructure. The beta is command-line first and runs on LangSmith Cloud in US regions only; further regions, an API and other deployment methods are promised later. No pricing was published.

What it means The hard part of shipping an agent has been the plumbing around it — durability, sandboxing, auth, scheduling. A hosted runtime that supplies all of it moves the build-or-buy line for a team without a platform group.