CrewAI — Multi-Agent Orchestration Framework

Core Concepts

Agents

Each agent has a role, goal, and optional tools:

analyst = Agent(
    role="Data Analyst",
    goal="Analyze datasets and find actionable insights",
    backstory="10 years of experience in data science",
    tools=[csv_tool, sql_tool, chart_tool],
    llm="claude-sonnet-4-20250514"
)

Tasks

Tasks define what needs to be done:

analysis_task = Task(
    description="Analyze Q1 sales data and identify top-performing regions",
    expected_output="Report with charts and recommendations",
    agent=analyst,
    output_file="q1_analysis.md"
)

Crews

Crews orchestrate agents and tasks:

crew = Crew(
    agents=[analyst, writer, reviewer],
    tasks=[analysis_task, write_task, review_task],
    process=Process.sequential  # or Process.hierarchical
)

Process Types

Built-In Tools

from crewai_tools import (
    SerperDevTool,      # Web search
    ScrapeWebsiteTool,  # Web scraping
    FileReadTool,       # Read files
    DirectoryReadTool,  # Read directories
    CodeInterpreterTool # Execute code
)

Real-World Example: Content Pipeline

# Research → Write → Edit → Publish
researcher = Agent(role="Researcher", tools=[search_tool])
writer = Agent(role="Writer")
editor = Agent(role="Editor")

crew = Crew(
    agents=[researcher, writer, editor],
    tasks=[
        Task(description="Research topic X", agent=researcher),
        Task(description="Write 1000-word article", agent=writer),
        Task(description="Edit for grammar and clarity", agent=editor),
    ],
    process=Process.sequential
)

Key Stats

• 25,000+ GitHub stars
• Role-based agent design
• Sequential and hierarchical processes
• 20+ built-in tools
• Memory and caching support

FAQ

Q: What is CrewAI? A: CrewAI is a Python framework for building teams of AI agents that collaborate on complex tasks, each with specialized roles, goals, and tools.

Q: Is CrewAI free? A: Yes, open-source under MIT license. Enterprise edition available.

Q: How is CrewAI different from LangGraph? A: CrewAI uses a role-playing metaphor — you define agents with roles and goals. LangGraph uses a graph-based approach with nodes and edges. CrewAI is more intuitive for team-like workflows.