Executive Summary
Built a conversational AI agent for a startup's hospital supply chain product. Healthcare practitioners were wasting time navigating systems and reports for simple answers like inventory levels, vendor options, or item substitutes. The AI agent using OpenAI and LangChain eliminated navigation entirely. Hospital staff can ask questions in plain language and get instant answers. The feature became a key differentiator in the competitive healthcare market.
Challenge
A startup building supply chain solutions for hospitals needed to differentiate its product in a competitive market. Its target users, healthcare practitioners, were wasting time navigating systems and digging through reports for simple supply chain answers: inventory levels, vendor options, and item substitutes.
The startup needed to make supply chain data instantly accessible through natural conversation, removing the friction of complex navigation entirely.
Approach
The real need wasn't a better dashboard or more reports, it was eliminating navigation entirely so hospital staff could get answers without extra steps.
Key considerations:
- Target users are time-constrained healthcare practitioners, not supply chain specialists
- Common queries are simple and predictable ("Do we have this in stock?", "What's an alternative?")
- Must integrate seamlessly with the startup's existing Firebase-based product architecture
- Natural language interface removes learning curve and adoption friction
Solution
I designed and built a conversational AI agent integrated into the startup's product with the following components:
- Natural language interface: Hospital staff query in plain language. No menus, no navigation, no training required
- Flask API backend: Integrated with Firebase Firestore for real-time querying of supplies, vendors, and surgery data
- RAG-powered AI agent: OpenAI GPT models with LangChain for retrieval-augmented generation, understanding queries and generating accurate responses from supply chain data
- Multiple data endpoints: Inventory checks, vendor analytics, waste tracking, and item substitution recommendations
- Visual analysis support: Generates charts and reports when staff need deeper exploration beyond simple queries
- Scalable deployment: Modular API architecture deployed via Heroku for easy integration and future expansion
Outcome
- Instant answers: Hospital staff get supply chain information in seconds via natural conversation
- Zero navigation: Eliminated need to dig through systems or wait for reports
- Product differentiation: Conversational AI became a core feature of the startup's offering in the competitive healthcare market
- Data-backed suggestions: The system provides alternatives and vendor comparisons based on actual supply chain data
- Seamless adoption: Natural language interface requires no training or learning curve
Tech Stack
Python • RAG (Retrieval-Augmented Generation) • AI Agent • Flask • Firebase Firestore • OpenAI API • LangChain • Pandas • Matplotlib • Heroku
