Building and Deploying Managed Agents (LangChain Academy)
by LangChain
LangChain's free course on taking a deep agent to production on LangSmith's managed runtime, with sandboxes, connections, schedules, auth, channels and evals.
Overview
Building and Deploying Managed Agents is a free LangChain Academy course that teaches the managed runtime LangChain launched in 2026 for its open-source Deep Agents harness. LangChain opened Managed Deep Agents as a private beta on May 13, 2026 and moved it to public beta on August 7, 2026, on LangSmith Cloud in the US region only and CLI-first while the API is finalized. The course has 24 lessons in four modules. Module 1, Building the Agent, covers a managed agents overview, your first managed agent, context, tools and middleware, and subagents, ending in a lab. Module 2, Running the Agent, covers the Agent Server, sandboxes, connections and schedules, with a lab. Module 3, Calling the Agent, covers identity, what a channel is, Slack channels, HTTP channels and building a custom frontend, with a third lab. Module 4 covers evals. According to LangChain's launch post, the runtime provides durable execution that can pause, retry and resume, persistent thread state, one sandbox per thread by default, memory under /memories/, OIDC identity that scopes threads per end user, managed cron schedules, LangSmith traces, Context Hub sync for instructions and skills, and evals through Harbor. Agents are written in Python or TypeScript and deployed with the mda CLI. This is vendor training for a specific product, so plan on a LangSmith account.
At a Glance
- Topic
- Agentic
- Level
- Intermediate
- Format
- Course
- Cost
- Free
- Duration
- 24 lessons across 4 modules, including 3 hands-on labs, self-paced (the page lists no video hours)
- Provider
- LangChain
- Hands-on
- Yes — code/exercises
- Certificate
- None
What You’ll Learn
- ✓Build a first managed agent and give it context, tools, middleware and subagents
- ✓Run an agent on the LangSmith Agent Server with per-thread sandboxes for code and files
- ✓Configure connections so a managed agent can reach external systems and services
- ✓Set up schedules so an agent runs recurring jobs without a human trigger
- ✓Scope agent threads to end users with identity and authentication
- ✓Expose an agent through Slack channels, HTTP channels and a custom frontend
- ✓Add evals that catch agent regressions before a change reaches production
Highlights
- •Covers the operations work that framework tutorials skip: sandboxing, scheduling, identity, delivery channels and evals for an agent that is already built
- •Three labs, one at the end of each of the first three modules, so you deploy and call a working agent instead of only reading about it
- •Tracks a runtime LangChain shipped in 2026 (public beta August 7, 2026), so the material is current to the product
- •Free, from the team that maintains Deep Agents, LangGraph and LangSmith
Who It’s For
Best For
- ✓Engineers who have a working Deep Agents or LangGraph agent and need to host it in production
- ✓Platform teams evaluating managed agent runtimes against self-hosting
- ✓Developers building agents that must serve many end users through Slack or an HTTP API
Prerequisites
- •Comfort writing agents in Python or TypeScript; LangChain's Introduction to Deep Agents course is a natural first step
- •A LangSmith Cloud account; Managed Deep Agents is in public beta in the US region only
- •Basic familiarity with the command line, since deployment is CLI-first
FAQ
What is Building and Deploying Managed Agents (LangChain Academy)?
Building and Deploying Managed Agents is a free LangChain Academy course for developers who already build agents and now need to run them in production. It walks through LangSmith's managed agent runtime: building an agent with context, tools, middleware and subagents, then running it with sandboxes and schedules, exposing it through identity-aware Slack and HTTP channels, and evaluating it.
Is Building and Deploying Managed Agents (LangChain Academy) free?
Building and Deploying Managed Agents (LangChain Academy) is free to access.
What level is Building and Deploying Managed Agents (LangChain Academy) for?
Building and Deploying Managed Agents (LangChain Academy) is aimed at a intermediate audience. Recommended background: Comfort writing agents in Python or TypeScript; LangChain's Introduction to Deep Agents course is a natural first step, A LangSmith Cloud account; Managed Deep Agents is in public beta in the US region only, Basic familiarity with the command line, since deployment is CLI-first.
How long does Building and Deploying Managed Agents (LangChain Academy) take?
Expect roughly 24 lessons across 4 modules, including 3 hands-on labs, self-paced (the page lists no video hours). Most learners work through it at their own pace.
What will I learn from Building and Deploying Managed Agents (LangChain Academy)?
You'll learn: Build a first managed agent and give it context, tools, middleware and subagents; Run an agent on the LangSmith Agent Server with per-thread sandboxes for code and files; Configure connections so a managed agent can reach external systems and services; Set up schedules so an agent runs recurring jobs without a human trigger; Scope agent threads to end users with identity and authentication; Expose an agent through Slack channels, HTTP channels and a custom frontend; Add evals that catch agent regressions before a change reaches production.
Topics
Sources
This page was written from 3 sources, 2 on domains other than academy.langchain.com.