What is crewAI?
It is an open-source framework that orchestrates role-playing AI agents, facilitating collaboration to seamlessly connect these AI agents to tackle complex problems.
Does that sound a bit stiff and hard to understand? Let me give you an example: “I want to output a technical article about AI.”
How do we do this using crewAI?
-
Define an “Advanced Research Analyst” agent, whose goal is to search and analyze the latest developments in AI and data science, with the main task being to conduct a comprehensive analysis of the latest AI advancements in 2024, identifying key trends, breakthrough technologies, and potential industry impacts, ultimately outputting a complete analysis report.
-
Define a “Technical Content Strategist” agent, whose goal is to write engaging content about technological advancements, with the main task being to use the provided analysis report to write an engaging blog post that highlights the most significant AI advancements, ensuring the article is informative yet easy to understand, suitable for a tech-savvy audience, avoiding complex vocabulary, and the final output must be a complete blog post of at least 4 paragraphs.
Let me share what I find interesting about this:
-
The output of the first agent can serve as the input for the second agent.
-
Each agent can choose a suitable large model based on its goals. For example, if model A is suitable for analysis and model B is suitable for writing, then the first agent can choose model A, while the second agent can choose model B.
-
Defining goals and executing tasks for agents is done using prompts, eliminating the need for complex coding.
Now, here’s the question:
-
I want to insert images into the article, what should I do? crewAI also provides configuration tools, allowing us to customize tools for implementation.
-
I want to use my private knowledge base, what should I do? crewAI also offers solutions, which I am currently exploring.
By now, you should have an understanding of what crewAI is, and feel free to reach out to me if you have questions ~
What Features Does crewAI Offer?
-
Role-based agent design: Customize specific roles, goals, and tools for agents.
-
Autonomous task delegation between agents: Agents can autonomously delegate tasks and inquire among themselves, improving problem-solving efficiency.
-
Flexible task management: Define tasks using customizable tools and dynamically assign them to agents.
-
Process-driven: Currently supports sequential task execution and hierarchical processes, with more complex processes like consensus and autonomous processes under development.
-
Save outputs as files: Save the outputs of individual tasks as files for later use.
-
Parse outputs into Pydantic or JSON: Parse the outputs of individual tasks into Pydantic models or JSON format.
-
Compatible with open-source models: Run your team using Open AI or open-source models or local models!
What Cases Does crewAI Provide?
-
Login page generator -
Incorporating human input during execution -
Travel planner -
Stock analysis
The Current Status of crewAI
The v0.1.0 version was released on 2023-11-15, and it has already received 9.6K stars on GitHub. The latest version is v0.22.5, which is updated quite frequently.
In Conclusion
crewAI provides a multi-agent collaboration platform that enhances the ability to solve complex problems through interactions and collaboration between agents, which is difficult for single-agent systems to match.
Personally, I believe that the products created by a multi-agent collaboration platform can disrupt many current single-agent applications. With proper task orchestration, there is great potential.
In the future, I will use the crewAI framework to implement some cases. Please stay tuned; I will open-source the code at: https://github.com/xinliangnote/crewAI-labs. If you want to discuss, you can add me on WeChat and join the discussion group.
Continuously sharing AIGC and large models, personal GitHub with 9K+ stars, welcome to follow.