Three implementation paths compare a direct model call, a LangChain agent built on LangGraph, and an explicit state graph with increasing orchestration control.
Free preview You are viewing part of LangChain & LangGraph Agents Workshop for free. Unlock the full course for $247
Three implementation paths compare a direct model call, a LangChain agent built on LangGraph, and an explicit state graph with increasing orchestration control.
Select the least complex abstraction that provides the control, state, and operating guarantees the application actually needs.

Establish a current, implementation-level mental model of LangChain agents, LangGraph orchestration, and project boundaries before learners introduce stateful complexity.

Module objectives

  • Distinguish LangChain high-level agent APIs, LangGraph orchestration/runtime primitives, and LangSmith managed services.
  • Implement a small bounded agent with current create_agent patterns and identify material Python/JavaScript differences.
  • Allocate responsibility among the model, instructions, middleware, runtime context, tools, deterministic code, and surrounding application.
  • Recognize when a direct model call or deterministic workflow is preferable to an agent loop.

Detailed topic content is generated and researched separately so each topic maintains its own source trail.

Recap slides

This module in 6 slides

Click through for a quick overview, or use your arrow keys.

Go deeper

Put this module to work

Suggested time: 3 to 4 hours

Real-world scenario

A team mixes old LangChain agent examples, new LangGraph code and LangSmith tracing without knowing which piece does what. Upgrades break things, and nobody can say whether a bug is in the chain, the graph or the tracing. This module sorts out the current roles: LangChain for models and building blocks, LangGraph for stateful control flow and LangSmith for tracing and evaluation, and when to use each.

Step-by-step walkthrough

  1. Install current versions and confirm them against the official documentation.
  2. Build a bounded agent using the current create_agent pattern with one tool and a clear system prompt.
  3. Turn on LangSmith tracing and inspect one run.
  4. Write down the agent’s boundaries: what it may do, what it must hand to code and what it must refuse.
  5. Identify any deterministic logic that should live outside the model.

Common mistakes and how to fix them

MistakeCopying deprecated examples.
FixCheck every example against the current docs. Older agent APIs have been replaced.
MistakeNo tracing from the start.
FixTurn tracing on from the first run. It makes every later step easier.
MistakeLetting the model handle fixed rules.
FixPut fixed rules in code or graph structure.
Daniel Morris

Field note from Daniel

I have seen many projects get stuck on versions and old tutorials. My habit, from years of maintaining live systems, is to pin versions, read the current docs and turn on logging before writing anything clever. It saves days.

Try this

Find an older LangChain tutorial and list every API in it that the current docs have replaced.

Before you move on

Your checklist is saved in this browser.

Interactive tool

Agent opportunity scorer

Rate your idea on five factors and get a recommendation: agent, plain workflow, or simplify first. Your work is saved in this browser.

Interactive tool

Autonomy envelope builder

List the actions your agent might take and sort each into Do alone, Ask first or Never. Your work is saved in this browser.

Like what you see? The full course goes much deeper.All modules, topics, interactive tools, knowledge checks and your certificate. Backed by a 30-day money-back guarantee.

Enroll now for $247