The 3-Agent Pattern: How Chex Built a Full-Stack Mental Health Assistant

Chex developed a novel three-agent system for mental health support that won a hackathon by effectively addressing the limitations of traditional AI assistants. Unlike typical chatbots, these agents are specialized and capable of performing distinct tasks such as booking therapy appointments, a function demonstrated through a live demo that impressed judges. The system consists of three agents: the Check-In Agent, which monitors wellness and initiates supportive interactions; the Resource Finder, which researches and compiles personalized lists of therapists; and the Scheduling Agent, which automates appointment bookings through calls and calendar management. This multi-agent architecture is characterized by its separation of concerns, tool-specific authentication, asynchronous operations, and specialized interfaces, allowing each agent to operate independently and efficiently. The approach highlights the effectiveness of decomposing complex workflows into focused agents, suggesting its applicability beyond mental health to fields like sales, DevOps, and recruiting, where similar structured and authenticated processes could be implemented.