crewAI Multi-Agent Production Systems

This talk will demonstrate how to design, build, and deploy a practical multi-agent system to automate and optimize sales and marketing lead generation and evaluating user generating content. We’ll move beyond theoretical concepts to address a critical business problems with actionable AI agent solutions.

We’ll begin by reverse-engineering the human process of identifying and qualifying leads, pinpointing the critical decision points and information gathering tasks. This analysis will guide us in defining the specific roles, responsibilities, and collaborative workflows for our AI agents.

Using frameworks like crewAI and PocketFlow, we’ll dive into the practical aspects of building these agents:

• Defining agentic tasks and tools.
• Crafting effective system and user prompts for complex, multi-step operations.
• Distinguishing between simple Python/Rust automation and true agentic orchestration that adapts and strategizes.

Finally, we’ll cover the essential steps for real-world deployment, including containerization on AWS and robust monitoring strategies leveraging tracing tools to provide crucial operational insights for developers. Attendees will leave with a clear blueprint for applying multi-agent AI to solve tangible business challenges.