Multi-Agent Systems with CrewAI#

CrewAI enables you to create teams of AI agents that work together autonomously. Each agent has a specific role, goal, and backstory that guides its behavior.

What is CrewAI?#

CrewAI is a framework for orchestrating role-playing autonomous agents. It allows you to:

Define agents with specific personas and expertise
Create tasks with clear objectives
Orchestrate collaboration between agents
Delegate work automatically based on agent capabilities

Installation#

bash

pip install crewai crewai-tools

Core Components#

Agents#

Agents are the building blocks of your crew:

python

from crewai import Agent

researcher = Agent(
    role="Senior Research Analyst",
    goal="Uncover cutting-edge developments in AI",
    backstory="""You work at a leading tech think tank.
    Your expertise lies in identifying emerging trends.""",
    tools=[search_tool, scrape_tool],
    llm=claude_3,
    verbose=True
)

writer = Agent(
    role="Tech Content Strategist",
    goal="Craft compelling content about AI discoveries",
    backstory="""You are a renowned content strategist
    known for making complex topics accessible.""",
    llm=claude_3,
    verbose=True
)

Tasks#

Tasks define what agents need to accomplish:

python

from crewai import Task

research_task = Task(
    description="""Conduct comprehensive research on the latest
    AI agent frameworks released in 2024.""",
    expected_output="A detailed report with key findings",
    agent=researcher
)

writing_task = Task(
    description="""Using the research findings, create a
    blog post about the top AI agent frameworks.""",
    expected_output="A polished blog post ready for publication",
    agent=writer,
    context=[research_task]  # This task depends on research
)

Crew Assembly#

python

from crewai import Crew, Process

crew = Crew(
    agents=[researcher, writer],
    tasks=[research_task, writing_task],
    process=Process.sequential,  # or Process.hierarchical
    verbose=True
)

result = crew.kickoff()

Advanced Patterns#

Hierarchical Process#

For complex workflows, use a manager agent:

python

from crewai import Crew, Process

crew = Crew(
    agents=[researcher, writer, editor],
    tasks=[research_task, writing_task, editing_task],
    process=Process.hierarchical,
    manager_llm=gpt4,  # Manager agent uses GPT-4
)

Custom Tools#

Create specialized tools for your agents:

python

from crewai_tools import BaseTool

class DataAnalysisTool(BaseTool):
    name: str = "Data Analyzer"
    description: str = "Analyzes datasets and returns insights"

    def _run(self, dataset_path: str) -> str:
        # Implementation
        return analysis_results

Best Practices#

Clear Role Definition: Give agents distinct, non-overlapping roles
Specific Goals: Make agent goals measurable and achievable
Rich Backstories: Detailed backstories improve agent reasoning
Task Dependencies: Use context to chain task outputs
Memory Management: Enable memory for long-running crews

Real-World Use Cases#

Content Creation Pipeline: ResearchWriting → Editing → SEO
Customer Support: TriageResolution → Follow-up
Data Analysis: CollectionProcessing → Visualization → Reporting