RAG-Based Insurance Policy Analyzer

An AI-powered full-stack web application that helps users understand their insurance policies through intelligent analysis, risk scoring, and a conversational chatbot — built with FastAPI, Next.js, and Groq's Llama 3.3.

Live Demo: rag-insurance-analyzer.vercel.app
Backend API Docs: rag-based-insurance-analyzer.onrender.com/docs

---

Why I Built This

Insurance policies are notoriously difficult to understand — filled with legal jargon, hidden exclusions, and complex terms. I built this tool to make policy analysis accessible to everyone. Upload your PDF, get a clean breakdown, and chat with an AI that actually knows your policy.

---

Features

• PDF Upload and Parsing — Upload any insurance policy PDF and get instant structured insights
• 5-Tab Results View — Coverage, Exclusions, Premiums, Claims, and Risk Score
• AI Risk Scoring — Overall risk score out of 10 with favorable and unfavorable aspects
• RAG Chatbot — Ask natural language questions, answered strictly from your policy data
• Chat Memory — Conversation history persisted per policy using localStorage
• User Dashboard — View and manage all your uploaded policies in one place
• Policy Delete — Remove old policies from your dashboard
• JWT Authentication — Secure login and signup with token-based auth

---

Architecture

---

User Flow

---

Tech Stack

Backend

• FastAPI
• SQLAlchemy + SQLite
• pdfplumber for PDF extraction
• Groq Llama 3.3-70B for AI parsing and RAG chat
• passlib + bcrypt for password hashing
• python-jose for JWT tokens

Frontend

• Next.js 14 with App Router
• Tailwind CSS
• localStorage for chat persistence

Deployment

• Vercel for frontend
• Render for backend

---

Project Structure

---

Local Setup

Prerequisites

• Python 3.11+
• Node.js 18+
• Groq API Key — free at console.groq.com

Backend

git clone https://github.com/rahulchand017/RAG_Based_Insurance_Analyzer.git
cd RAG_Based_Insurance_Analyzer

python -m venv venv
venv\Scripts\Activate.ps1   # Windows
source venv/bin/activate     # Mac/Linux

pip install -r requirements.txt

Create a .env file:

uvicorn main:app --reload
# Runs at http://localhost:8000
# API docs at http://localhost:8000/docs

Frontend

cd frontend
npm install

Create a .env.local file:

npm run dev
# Runs at http://localhost:3000

---

API Endpoints

Method	Endpoint	Description
POST	/register	Create account
POST	/login	Login and get token
POST	/upload-policy	Upload PDF
POST	/analyze-policy/{id}	Run AI analysis
GET	/policy/{id}	Get full policy data
GET	/my-policies	Get user's policies
DELETE	/policy/{id}	Delete a policy
POST	/chat	Ask question via RAG

---

How RAG Works

1. User asks a question about their policy
2. Backend fetches all structured data for that policy from the database
3. The data is injected as context into the Groq prompt
4. Llama 3.3-70B answers based strictly on the policy — no hallucinations
5. Response is returned to the user in the chat UI

---

Contact

Rahul Chand
GitHub: @rahulchand017

---

If you found this useful, consider giving it a star on GitHub.