CrewAI Inc.

CrewAI

Role-playing multi-agent teams.

CrewAI orchestrates multiple LLM agents that play defined roles and collaborate on tasks. A lean alternative to heavier agent frameworks when you want a small team of specialists working a structured plan.

Install

bash
pip install crewai

Quickstart

A minimal example to verify your setup.

python
from crewai import Agent, Task, Crew

researcher = Agent(role="Researcher", goal="Find facts", backstory="PhD analyst")
writer = Agent(role="Writer", goal="Explain clearly", backstory="Tech writer")

task = Task(
    description="Summarize the state of RAG in 2025",
    expected_output="3 bullet summary",
    agent=writer,
)

crew = Crew(agents=[researcher, writer], tasks=[task])
print(crew.kickoff())

Core concepts

Agents

Each agent has a role, goal, backstory, and toolset — small biographies that steer behavior.

Tasks

Discrete units of work with expected outputs, assigned to agents and chained sequentially or hierarchically.

Crews & processes

Sequential or hierarchical processes coordinate the team, with a manager agent in the latter.

Flows

Event-driven control plane for branching, loops, and conditional crew runs.

Common use cases

  • Research and reporting pipelines
  • Content generation with editor review
  • Sales prospecting agents
  • Multi-step ops automations

Resources