CrewAI

Multi-agent · by CrewAI

CrewAI

Give each agent a role, a goal and a backstory, then let the crew work the task.

35 lessons3h 35m reading6 sections

01Your first crew0/8 · 42m

Introduction to CrewAI3m

Setup: installing CrewAI4m

A support desk in plain Python4m

A custom LLM with BaseLLM7m

Your first crew: agents and tasks7m

The agent prompt: what the model is sent6m

Inputs: one crew, many tickets5m

Watching a crew work6m

Section quiz · 4 questionsNot taken

02Tools and hooks0/6 · 37m

Tools: functions an agent can use5m

Tool calls: asking for a tool8m

Tool failures mid-run6m

max_iter: capping an agent's loop5m

Tool hooks: approving a refund7m

Model hooks around the call6m

Section quiz · 5 questionsNot taken

03Tasks and teams0/6 · 37m

Sequential tasks: two agents in a row7m

Structured output with output_pydantic6m

Guardrails: checking a task's answer6m

Conditional tasks that run only sometimes5m

Delegation between agents7m

Hierarchical process: a manager for the crew6m

Section quiz · 5 questionsNot taken

04Flows0/6 · 38m

First flow: steps in plain Python6m

Flow state with a Pydantic model5m

Router: choosing a path in a flow6m

Crew in a flow: one step of a flow6m

Human feedback: pausing a flow8m

Streaming a crew's output7m

Section quiz · 4 questionsNot taken

05Knowledge, memory and production0/5 · 35m

Knowledge sources: a reference library for agents8m

Memory: remember and recall across runs8m

MCP server tools for an agent7m

Real model: swapping in a hosted LLM5m

Integrations: from stand-in to production7m

Section quiz · 4 questionsNot taken

06The support desk0/4 · 26m

Desk state and a reply guardrail6m

Desk crew inside a flow7m

Three tickets through the desk6m

Desk refund: pausing for a manager7m

Section quiz · 3 questionsNot taken