Overview

CrewAI is a Python framework for orchestrating autonomous AI agents, allowing you to form virtual teams where each agent has specific roles and goals to collaboratively handle complex tasks.

Core Concepts

1. Agent

Agent is an AI execution unit with a specific role:

from crewai import Agent

researcher = Agent(
    role="Research Analyst",
    goal="Provide accurate, in-depth research analysis",
    backstory="You are a senior research analyst skilled at multi-perspective analysis.",
    verbose=True
)

2. Task

Task is specific work assigned to an agent:

from crewai import Task

research_task = Task(
    description="Analyze AI application trends in healthcare",
    agent=researcher,
    expected_output="A structured analysis report"
)

3. Crew

Crew coordinates multiple agents to complete tasks:

from crewai import Crew

crew = Crew(
    agents=[researcher, writer],
    tasks=[research_task, write_task],
    process="sequential"  # or "parallel"
)

Complete Example

from crewai import Agent, Task, Crew
from crewai.process import Process

# 1. Define Agents
researcher = Agent(
    role="Market Researcher",
    goal="Collect and analyze market data",
    backstory="You are a professional market analyst skilled in data analysis."
)

writer = Agent(
    role="Content Writer",
    goal="Write clear, professional market reports",
    backstory="You are a senior content writer who excels at transforming complex information into readable articles."
)

# 2. Define Tasks
research_task = Task(
    description="Collect 2024 AI industry market size and growth trend data",
    agent=researcher,
    expected_output="Market size data, growth rates, key players list"
)

write_task = Task(
    description="Write market analysis report based on research data",
    agent=writer,
    expected_output="A complete market analysis report with executive summary and conclusions"
)
write_task.context = [research_task]  # Depends on research_task

# 3. Create Crew
crew = Crew(
    agents=[researcher, writer],
    tasks=[research_task, write_task],
    process=Process.sequential
)

# 4. Execute
result = crew.kickoff()
print(result)

Process Configuration

Sequential

crew = Crew(
    agents=[agent1, agent2, agent3],
    tasks=[task1, task2, task3],
    process=Process.sequential
)
# Tasks execute in defined order

Hierarchical

crew = Crew(
    agents=[researcher1, researcher2, researcher3],
    tasks=[task1, task2, task3],
    process=Process.hierarchical
)

Common Questions

Q1: What is the context parameter for Task?

• Specifies prerequisite tasks for the current task
• Previous task outputs become context for the current task
• Used to establish dependencies between tasks

Q2: What's the difference between sequential and hierarchical?

• sequential: Tasks execute in order, each completes before the next starts
• hierarchical: A Manager Agent coordinates other agents

Q3: How to make agents collaborate?

• Establish dependencies via Task's context parameter
• Previous agent's output becomes next agent's input
• Shared Crew-level context

References

• CrewAI Official Documentation
• CrewAI GitHub