Introduction
In the previous part, we explored how to build structured AI workflows using CrewAI Flows.
Avi Chawla
We covered state management, flow control, and integrating a Crew into a flow.
As discussed last time, with Flows, you can create structured, event-driven workflows that seamlessly connect multiple tasks, manage state, and control the flow of execution in your AI applications.
This allows for the design and implementation of multi-step processes that leverage the full potential of CrewAI's capabilities.
However, in real-world applications, a single Crew is often not enough.
As we shall see shortly, complex workflows require multiple specialized Crews working together to complete different aspects of a larger process.
Thus, in this part, we shall focus on multi-crew flows, where several Crews operate in parallel or sequentially to achieve a common goal. We will walk through real-world use cases, breaking down how different Crews collaborate, share information, and optimize workflow execution.
By the end of this part, you will understand how to:
- Design and implement flows that combine multiple Crews efficiently.
- Use Crew dependencies to control task execution and coordination.
And of course, everything will be supported with proper implementations like we always do!
Let’s begin by recapping the fundamental principles of multi-crew architectures and their advantages.
If you haven't read Part 1, Part 2 and Part 3 yet, it is highly recommended to do so before moving ahead.
Avi Chawla
Avi Chawla
Avi ChawlaRecap
Why Flows?
As we learned in the previous parts, in traditional software development, workflows are meticulously crafted with explicit, deterministic logic to ensure predictable outcomes.
- IF A happens → Do X
- IF B happens → Do Y
- Else → Do Z
Of course, there's nothing wrong, and this approach excels in scenarios where tasks are well-defined and require precise control.
However, as we integrate Large Language Models (LLMs) into our systems, we encounter tasks that benefit from the LLMs' ability to reason, interpret context, and handle ambiguity—capabilities that deterministic logic alone cannot provide.
For instance, consider a customer support system.
Traditional logic can efficiently route queries based on keywords, but understanding nuanced customer sentiments or providing personalized responses requires the interpretative capabilities of LLMs.
Nonetheless, allowing LLMs to operate without constraints can lead to unpredictable behavior at times.
Therefore, it's crucial to balance structured workflows with AI-driven autonomy.
Flows help us do that.
Essentially, Flows enable developers to design workflows that seamlessly integrate deterministic processes with AI's adaptive reasoning.
By structuring interactions between traditional code and LLMs, Flows ensure that while AI agents have the autonomy to interpret and respond to complex inputs, they do so within a controlled and predictable framework.
In essence, Flows provide the infrastructure to harness the strengths of both traditional software logic and AI autonomy, creating cohesive systems that are both reliable and intelligent.
It will become easier to understand them once we get into the implementation so let's jump to that now!
Installation
Throughout this crash course, we have been using CrewAI, an open-source framework that makes it seamless to orchestrate role-playing, set goals, integrate tools, bring any of the popular LLMs, etc., to build autonomous AI agents.
crewAIInc
To highlight more, CrewAI is a standalone independent framework without any dependencies on Langchain or other agent frameworks.
Let's dive in!
To get started, install CrewAI as follows:
Like the RAG crash course, we shall be using Ollama to serve LLMs locally. That said, CrewAI integrates with several LLM providers like:
- OpenAI
- Gemini
- Groq
- Azure
- Fireworks AI
- Cerebras
- SambaNova
- and many more.
To set up OpenAI, create a .env file in the current directory and specify your OpenAI API key as follows:
Also, here's a step-by-step guide on using Ollama:
- Go to Ollama.com, select your operating system, and follow the instructions.
- If you are using Linux, you can run the following command:
- Ollama supports a bunch of models that are also listed in the model library:
Once you've found the model you're looking for, run this command in your terminal:
The above command will download the model locally, so give it some time to complete. But once it's done, you'll have Llama 3.2 3B running locally, as shown below which depicts Microsoft's Phi-3 served locally through Ollama:
That said, for our demo, we would be running Llama 3.2 1B model instead since it's smaller and will not take much memory:
Done!
Everything is set up now and we can move on to building our Flows.
Technical recap
Here's a brief recap of the technical details we discussed last time.
Also, you can download the code for this article below:
Read the full article
Sign up now to read the full article and get access to all articles for paying subscribers only.
Join today!
