In this hands-on workshop, participants will learn how to redesign project work across people and AI by creating a practical multi-agent system.

The session opens with the real story of how an 18-person volunteer team delivered 667 classroom packs containing 20,000 STEM and SEL resources within 30 days and for less than $500. The project was completed through human leadership, creativity, and disciplined coordination—not through an agentic system.

Participants will then use the completed project as a design challenge. They will identify which activities should remain human-led, which could be AI-assisted or delegated, and where approvals and escalation points are required. Working from the project requirements, they will design five specialized agents supporting operations, finance and inventory, scheduling, communications, and quality and risk.

During a guided build-along, participants will configure a role-based Custom GPT to support the five-agent system. They will define agent responsibilities, project knowledge, decision boundaries, expected outputs, and human oversight requirements.

Participants will leave with a Custom GPT prototype, reusable design tools, and a practical method for applying multi-agent systems to their own projects.

Key Takeaways

1. Identify and classify project activities as human-led, AI-assisted, or appropriate for delegation to an AI agent
2. Build a role-based Custom GPT prototype that coordinates specialized project agents and can be adapted to support their own projects
3. Design a multi-agent project system with clearly defined roles, responsibilities, decision boundaries, outputs, and escalation points