How to beginner · 3 min read

How to run a crew in CrewAI

Q: How to run a crew in CrewAI

To run a crew in CrewAI, you instantiate a Crew object with your agents, then call its run() method to execute the coordinated tasks. This manages multiple AI agents working together on complex workflows using simple Python code.

Quick answer

To run a crew in CrewAI, you instantiate a Crew object with your agents, then call its run() method to execute the coordinated tasks. This manages multiple AI agents working together on complex workflows using simple Python code.

PREREQUISITES

Python 3.8+
OpenAI API key (free tier works)
pip install crewai>=0.1

Setup

Install the crewai Python package and set your OpenAI API key as an environment variable.

Run pip install crewai>=0.1 to install.
Set your API key in your shell: export OPENAI_API_KEY='your_api_key' (Linux/macOS) or setx OPENAI_API_KEY "your_api_key" (Windows).

bash

pip install crewai>=0.1

Step by step

This example creates two simple agents and runs them as a crew to complete a task collaboratively.

python

import os
from crewai import Crew, Agent

# Define two simple agents
class AgentA(Agent):
    def run(self, input_text):
        return f"AgentA processed: {input_text}"

class AgentB(Agent):
    def run(self, input_text):
        return f"AgentB enhanced: {input_text.upper()}"

# Instantiate the crew with the agents
crew = Crew(agents=[AgentA(), AgentB()])

# Run the crew with an input
result = crew.run("Hello CrewAI")
print(result)

output

['AgentA processed: Hello CrewAI', 'AgentB enhanced: HELLO CREWAI']

Common variations

You can run crews asynchronously, use different AI models per agent, or stream outputs for real-time interaction.

python

import asyncio
from crewai import Crew, Agent

class AsyncAgent(Agent):
    async def run(self, input_text):
        await asyncio.sleep(1)  # simulate async work
        return f"AsyncAgent got: {input_text}"

async def main():
    crew = Crew(agents=[AsyncAgent()])
    result = await crew.run_async("Async input")
    print(result)

asyncio.run(main())

output

['AsyncAgent got: Async input']

Troubleshooting

If you see errors about missing API keys, ensure OPENAI_API_KEY is set in your environment. For timeout errors, increase agent timeout settings or check network connectivity.

✅

Key Takeaways

Use the Crew class to coordinate multiple agents easily in CrewAI.
Define agents by subclassing Agent and implementing the run() method.
Run crews synchronously or asynchronously depending on your workflow needs.
Always set your OpenAI API key in the environment to avoid authentication errors.
CrewAI supports flexible agent orchestration for complex multi-agent AI workflows.

Verified 2026-04 · gpt-4o, claude-3-5-sonnet-20241022

Verify ↗