Building with Agents

1 certified

Build agents that actually work in production. This course goes from picking a framework (LangChain vs LangGraph vs CrewAI) to shipping a stateful, observable, multi-step agent. Code-along tutorials, framework comparisons, and Anthropic's principles for effective agents.

✨Try the interactive lessons→

Certificate requires 10+ karma · you have 0

01
Start here: what this course is about
Welcome to Building with Agents
Read first. ~4 minutes.
02
Start here: Anthropic on building effective agents
Anthropic: Building Effective AI Agents↗
The principles that define quality agents. Anthropic's authoritative paper. Read this before picking any framework.
03
Compare the major frameworks (neutral)
AI Agent Frameworks Compared: LangGraph vs CrewAI vs AutoGen (2026)↗
Honest comparison: CrewAI (easiest), AutoGen (medium), LangGraph (most control). Pick what fits your problem, not what marketing says.
04
Anthropic Agent SDK (official)
Claude Agent SDK Overview - Anthropic↗
Build agents in Python or TypeScript with Anthropic's official SDK. Read files, run commands, search the web, edit code.
05
OpenAI Agents SDK (official)
OpenAI Agents SDK Guide↗
OpenAI's lightweight Python framework. Agents that plan, use tools, execute multi-step tasks. The official quickstart.
06
Hands-on: your first agent with LangChain
Build AI Agents with LangChain: Complete Developer Guide↗
Independent quality tutorial. Install, configure, run. The fastest path to a working agent without vendor bias.
07
Adding RAG to your agent
Build a RAG agent with LangChain - Official Docs↗
LangChain's official tutorial for RAG agents. Connect your agent to a knowledge base.
08
Build an agent from scratch (no framework)
A Super Simple ReAct Agent from Scratch↗
Quality tutorial. 50 lines of Python. Understand the loop before you let a framework hide it from you.
09
From toy to production
Building Production ReAct Agents From Scratch Is Simple↗
What changes between a demo agent and one that runs in production: errors, retries, observability, costs.
10
LangGraph for production (source)
GitHub - langchain-ai/langgraph: Build resilient agents↗
Official LangGraph repo. Stateful, durable, checkpointable agents. Read examples and patterns directly from source.
11
State of the art: MCP, A2A, multi-agent
MCP & Multi-Agent AI: Building Collaborative Intelligence 2026↗
Where the field is in 2026: MCP for tools, A2A for agent-to-agent, multi-agent orchestration as the new standard.
12
Closing: what to take with you
Closing: Frameworks Are Not the Skill
Wrap-up. ~3 minutes.