AI For Data Scientists: Multi-Agent Test

 

 

 

 

 

I've been tinkering with a new AI framework for making AI agents.

(It's not LangChain) A thread.

1. The framework is CrewAI.

CrewAI is a Python package.

The benefit of CrewAI is that you can build AI agents that operate in a cohesive unit with less code than LangChain and LangGraph.

Here's a quick example:

2. Setting up a CrewAI Project

In one line of code, CrewAI creates an AI project template with the necessary files to build and orchestrate a multi-agent system.

3. Agents.yml File

The agents.yaml file contains instructions to set up AI agents.

Here's a quick example:

Note - You can get more complex with RAG, function calling, and specialized tools like OpenAI's Data Explorer.

4. Tasks.yml file

The tasks.yml file is where you specify the tasks for your multi-agent project and you assign agents to each task.

5. Crew.py file

This is where you specify the structure of your crew.

6. Main.py file

This runs your AI Agent Workflow.

Learn more about CrewAI here: https://github.com/crewAIInc/crewAI

P.S. - Do you want to become a Generative AI Data Scientist?

On December 18th, I'm sharing one of my best Generative AI Projects: A SQL-Writing Business Intelligence Agent

Register here (500 seats): https://learn.business-science.io/ai-register