Kick off the workshop with a clear overview of what CrewAI is and why it's a game-changer in the realm of agent-based systems. We'll discuss the motivation behind multi-agent orchestration, walk through CrewAI’s core and advanced components (agents, tasks, tools, crews, flows), and map these to real-world business use cases such as content generation, fraud detection, customer service, and onboarding. You’ll also get an outline of the workshop structure and what to expect in each module.

• Why multi-agent systems?
• Core components: agents, tasks, crews, tools, flows
• Advanced features: conditional logic, async orchestration, LLM-agnostic setups
• Business case examples and architectural patterns

In this hands-on session, you’ll set up your environment and build your very first CrewAI application. Learn by doing: define a basic agent, assign it a task, and compose a simple crew. This module ensures all participants are equipped with working installations and a clear understanding of the basics.

• Environment setup (virtualenv/conda, dependencies)
• Installing and configuring CrewAI
• Creating your first agent and task
• Forming your first crew and running a workflow

Here, you'll build a content creation application using agents and flows, with built-in "guardrails" for maintaining quality, relevance, or tone. Learn how to define conditional orchestration strategies to guide the process across multiple agents, each responsible for different steps in content generation.

• Using flow for conditional task execution
• Implementing content generation with validation steps
• Guardrails through intermediate agents
• Real-world use case: blog/social media content generation

Design a fraud detection pipeline using agents trained to analyze transaction logs and behaviors. This scenario introduces a more data-driven approach, integrating tools and heuristics for pattern recognition and classification.

• Parsing and structuring transaction data
• Agents for anomaly detection and classification
• Using external tools (e.g., regex, statistical tests)
• Workflow coordination and response generation

Create an AI-powered onboarding system that adapts to users through personalized interactions. Leverage persistent memory (via mem0) to build context-aware conversations across sessions and allow agents to learn and recall user preferences.

• Designing agent memory with mem0
• Personalized task flow for onboarding
• Integration with user input tools (forms, questionnaires)
• Memory persistence strategies for long-term agent behavior

Wrap up the workshop with a capstone project: a multi-agent interview simulator with Streamlit. This end-to-end application connects CrewAI agents to a front-end UI and includes voice-to-text and text-to-speech capabilities for a seamless conversational experience. This module ties together everything learned so far and provides a production-style template for real-world applications.

• Creating an interactive front-end with Streamlit
• Setting up voice interfaces (text-to-speech, speech-to-text)
• Orchestrating a crew for mock interviews
• Connecting UI, tools, and CrewAI agents into a coherent system