Projects
Hospital Operations AI Assistant
This module presents an AI-powered assistant designed to support hospital operations at Northwestern Memorial Hospital. Built with LangChain, LangGraph, and GPT-4o-mini, the agent handles multiple roles including triaging cases, analyzing operational metrics, and coordinating recommendations across specialist tools. It simulates a supervisory persona (“Claire”) that routes tasks to sub-agents and synthesizes responses in plain language.
The project showcases how Retrieval-Augmented Generation (RAG), agent reasoning, and memory persistence can be combined to support complex workflows in regulated, high-stakes environments like healthcare, while preserving traceability and explainability in every step.
⸻
Features
- Role-Based Agent Design: Claire delegates tasks to specialist agents (e.g., Financial Planner, Utilization Review).
- Tool Integration: Connects to retrieval tools, structured memory, and analytical sub-systems for reporting and triage.
- Workflow Orchestration: Uses LangGraph to route and track multi-step interactions across conversation threads.
- Traceable Decisions: All recommendations are linked to specific inputs and model responses for transparency.
- Human-Centric Output: Final outputs are formatted for hospital leadership and frontline staff alike, emphasizing clarity and actionability.
⸻
Key Insight
AI systems are most valuable in high-stakes domains when they are modular, auditable, and human-facing. This project illustrates how LangGraph agents can be used not just to answer questions, but to assist with operational decisions, workload delegation, and contextual reasoning in complex organizational settings.