By Gaurab Chhetri - Hackathon Project
MedGuide AI - Hackathon Winning AI-Powered Medical Assistant
MedGuide AI is a privacy-first AI assistant that interprets medical documents and prescriptions using LangGraph, RAG, and agentic workflows. Built in 48 hours, it won the Organized AI Hackathon 2025 at Antler’s space.
MedGuide AI is an intelligent medical assistant built during the Organized AI Hackathon 2025 (hosted at Antler’s Austin space), where it won first place.
The tool empowers users to upload lab reports, prescriptions, or clinical summaries and receive clear, plain-language explanations powered by LangGraph, RAG (Retrieval-Augmented Generation), and agentic workflows. Privacy was a core principle: all processing happens locally, ensuring sensitive medical data never leaves the user’s environment.
Hackathon Highlights
- 🏆 Winner of Organized AI Hackathon 2025
- ⏱️ Built in under 48 hours
- 💡 Focused on healthcare accessibility and trust through AI
- 💻 Tech stack: Python, Streamlit, LangGraph, ChromaDB, MCP servers
Key Features
- Lab Test Interpretation: Upload PDFs of blood work or diagnostics and receive AI-generated insights
- Medication Guidance: Explain prescriptions like “Metformin 500mg BID” in plain language
- Document Memory: Persistent recall of past documents for continuity across sessions
- Privacy-First: Local vector storage and offline-capable AI workflows
Why It Matters
Healthcare data is often overwhelming and inaccessible. From cryptic lab values to confusing prescription shorthand, patients are left in the dark. MedGuide AI bridges this gap by combining agentic reasoning with retrieval and context-awareness, giving users safe, clear, and trustworthy explanations of their own health records.
Team
Built with an amazing team of collaborators: Samar Ranjit 1, Diwas Pandit 2, Sujal Pandey 3, and mentorship from the AITX community.