How to Build an AI Agent with CrewAI

Introduction

Learn how to build multi-agent systems using CrewAI framework from scratch.

Prerequisites

• Python 3.10+
• OpenAI API key (or compatible LLM API)
• Basic understanding of Python

Step 1: Installation

pip install crewai

Step 2: Set Up Your API Key

export OPENAI_API_KEY="your-api-key-here"

Step 3: Define Your Agents

from crewai import Agent

# Research agent
researcher = Agent(
    role='Senior Research Analyst',
    goal='Uncover cutting-edge developments in AI and data science',
    backstory="""You are an expert at understanding the nuances of AI and data science.
    You are skilled at turning complex data into actionable insights.""",
    verbose=True,
    allow_delegation=True,
)

# Writer agent
writer = Agent(
    role='Technical Writer',
    goal='Create well-structured, engaging technical content',
    backstory="""You are a skilled technical writer with expertise in AI and technology.
    You can explain complex topics in an accessible way.""",
    verbose=True,
    allow_delegation=False,
)

Step 4: Define Your Tasks

from crewai import Task

research_task = Task(
    description="""
        Analyze the latest developments in AI and data science.
        Focus on:
        - Recent breakthroughs
        - Emerging trends
        - Industry impact
    """,
    expected_output="A comprehensive research report on AI developments",
    agent=researcher,
)

writing_task = Task(
    description="""
        Write a technical blog post based on the research findings.
        Make it accessible to developers and technical enthusiasts.
    """,
    expected_output="A well-structured blog post ready for publication",
    agent=writer,
    context=[research_task],
)

Step 5: Create Your Crew and Run

from crewai import Crew

crew = Crew(
    agents=[researcher, writer],
    tasks=[research_task, writing_task],
    verbose=2,
)

result = crew.kickoff()
print(result)

Advanced: Custom Tools

Create a Custom Tool

from crewai import Agent, Task, Crew, Tool
from crewai_tools import SerperDevTool

# Use built-in search tool
search_tool = SerperDevTool()

researcher = Agent(
    role='Research Analyst',
    goal='Find the latest information',
    backstory='Expert researcher',
    tools=[search_tool],
)

Create a Custom Tool

from crewai_tools import BaseTool

class MyCustomTool(BaseTool):
    name: str = "My Custom Tool"
    description: str = "Does something useful"

    def _run(self, argument: str) -> str:
        # Your custom logic here
        return f"Result for: {argument}"

tool = MyCustomTool()
agent = Agent(role='Agent', goal='Goal', backstory='Backstory', tools=[tool])

Process Types

Process	Description
sequential	Tasks execute in order (default).
hierarchical	Manager agent assigns tasks.
consensual	All agents agree on assignments.

Using Hierarchical Process

crew = Crew(
    agents=[researcher, writer],
    tasks=[research_task, writing_task],
    process=Process.hierarchical,
    manager_llm="gpt-4",
)

Best Practices

1. Define clear roles - Each agent should have a specific purpose.
2. Write detailed backstories - Helps guide agent behavior.
3. Use task context - Pass outputs between tasks.
4. Enable delegation - Let agents collaborate.
5. Use verbose mode - For debugging and understanding.

Troubleshooting

Issue	Solution
API errors	Check your API key and quota
Poor results	Improve agent backstories and task descriptions
Slow execution	Reduce verbose mode, use smaller models for simple tasks
Delegation loops	Set allow_delegation=False on critical agents

Resources

• CrewAI Documentation
• GitHub Repository
• Examples Gallery