🤖 HackFusionRx AI Assistant

The AI Assistant is the core intelligence layer of HackFusionRx, responsible for converting prescriptions into structured, bill-ready data.

It supports voice, text, and handwritten prescription inputs.

---

🚀 Features

🗣️ Voice & Text Input

• Accepts spoken or typed prescriptions
• Converts to structured JSON using LLM

---

🧾 Prescription Parsing

• Extracts:

  • Patient name
  • Medicine name
  • Dosage
  • Frequency
  • Quantity

---

✍️ Handwritten OCR

• Upload prescription image
• Uses Claude Vision for text extraction
• Passes result to LLM pipeline

---

🧠 Smart Matching

• Fuzzy matching for:

  • Customers
  • Medicines

• Handles variations like:

  • "Paracetamol" vs "Paracetamol 500mg"

---

💳 Bill Generation

• Auto-generates bill UI

• Shows:

  • Stock availability
  • Pricing
  • Shortages

---

📦 Order Processing

• Creates order in Firestore
• Creates transaction in RTDB
• Updates stock automatically
• Updates customer stats

---

📧 Email Integration

• Sends styled HTML bill
• Uses Gmail SMTP via Nodemailer

---

🧠 AI Pipeline

Input (Voice/Text/Image)
        ↓
Claude Vision (OCR)
        ↓
Ollama (LLaMA 3.2)
        ↓
Structured JSON
        ↓
Database Matching
        ↓
Bill Generation

---

🏗️ Tech Stack

• React (Vite)
• Node.js + Express
• Ollama (LLaMA 3.2)
• Claude Vision API
• Firebase (Firestore + RTDB)

---

⚙️ API Endpoints

• /api/extract-prescription → LLM parsing
• /api/ocr-prescription → OCR processing
• /api/create-order → Save order
• /api/send-bill → Email bill

---

🛠️ Setup Instructions

1. Install dependencies

npm install

---

2. Run AI Assistant

npm run dev

Runs on: http://localhost:5173

---

⚠️ IMPORTANT

👉 The AI Assistant will NOT function independently.

You MUST run:

• Backend (port 3001)
• Main frontend (port 5174)
• AI Assistant (port 5173)

👉 All three services must run simultaneously.

---

🔑 Requirements

• Ollama running locally
• Claude API key configured
• Firebase setup completed

---

🌟 Key Highlights

• End-to-end prescription automation
• OCR + LLM hybrid pipeline
• Real-time stock-aware billing
• Minimal manual input required

---

📌 Future Enhancements

• Multi-language prescription support
• Drug interaction warnings
• Voice-to-bill fully automated flow

---