Kicking Off Crews with Advanced Functionalities

Imagine you're the leader of a specialized team, ready to launch a complex mission. Your team is like a well-coordinated group of agents, each with a specific role to play in tackling a challenging task. Now, picture this: What if you need to launch that mission not just once, but multiple times—maybe for different pieces of information? Or what if you need your team to keep working in the background without any supervision while you focus on other important tasks?

In this lesson, we'll explore these advanced strategies—how to set your team in motion for every item in a list, how to keep them working asynchronously so they don’t need to wait for each other, and how to replay a mission from a specific point if needed. By the end, you’ll be able to use CrewAI's multi-agent systems more dynamically and flexibly, allowing you to easily handle complex agentic workflows. This will empower you to get the most out of your "crew," no matter how tricky the mission.

How to have multiple inputs for the Crew's kickoff

In real-world projects, tasks often come in groups, not just one at a time. Think about being asked to analyze the ages of participants from different groups. Instead of handling each group manually, wouldn’t it be more efficient to have a system that does it all at once? That’s where CrewAI comes in. With CrewAI, you can launch a crew to automatically handle each item in a list. It’s like deploying your team to tackle multiple challenges simultaneously, each team member focusing on a different dataset but performing the same analysis.

Imagine you’ve got several datasets and want to analyze the ages of participants in each one as we did in the previous lesson. Instead of running the analysis manually for each dataset, CrewAI automates the process. To accomplish this, CrewAI provides the kickoff_for_each() method. This method is designed to efficiently execute the same crew for each item in a list.

Note: The description of agents and tasks are skipped in the explanation for this lesson so you can focus on the topic at hand. However, to see everything in detail, you can check out the Jupyter Notebooks.