AI-Powered Hospital Appointment Management Platform

A mid-sized hospital network managing multiple specialties relied on heavily manual appointment processes — phone calls, paper-based logs, and inconsistent scheduling practices across departments. Patients faced long wait times and scheduling conflicts, while repeated cancellations and last-minute reschedules were nearly impossible to track systematically. The hospital had no centralised system for automated reminders, conflict resolution, or reporting. This drove up administrative workload for reception and staff, caused revenue loss from missed appointments, led to inefficient allocation of doctors and lab resources, and left mobile-first patients with a frustrating, friction-heavy experience that traditional chatbots could not solve.

We designed and implemented a hierarchical multi-agent AI system orchestrated using LangGraph, deployed on a fully cloud-native AWS stack. A main Supervisor Agent receives all patient requests and routes them to two specialised supervisor agents — one for doctor management and one for lab test management — each of which delegates to individual worker agents responsible for availability lookups and booking actions. The FastAPI backend runs in Docker containers on AWS ECS Fargate, accessed via API Gateway and a Load Balancer. Conversation memory is persisted in Amazon S3, allowing agents to maintain context across sessions. The Next.js frontend is hosted on AWS Amplify with CI/CD integration, and all data is stored in Amazon RDS PostgreSQL. An AI scheduling optimiser handles multi-resource allocation (doctor + lab + room) per appointment, and automated reminders are dispatched via AWS SNS.