🛡️ Insurance AI Copilot

An AI-powered insurance premium prediction and intelligent policy assistant built using Machine Learning, FastAPI, Retrieval-Augmented Generation (RAG), Google Gemini, and ChromaDB.

---

📌 Overview

Insurance AI Copilot is an intelligent web application that combines Machine Learning with Large Language Models (LLMs) to modernize insurance services.

The application predicts insurance premiums using a trained LightGBM model while simultaneously providing an AI-powered chatbot capable of answering insurance-related questions using Retrieval-Augmented Generation (RAG).

Unlike traditional chatbots, responses are generated from a dedicated knowledge base, producing more reliable and context-aware answers.

---

✨ Features

🤖 AI Insurance Assistant

• Gemini-powered conversational assistant
• Context-aware responses using RAG
• Insurance-specific knowledge retrieval
• Natural language interactions

📈 Premium Prediction

• Predict insurance premium instantly
• Machine Learning powered
• LightGBM prediction model
• Fast inference

🧠 Retrieval-Augmented Generation (RAG)

• ChromaDB vector database
• Semantic document retrieval
• Grounded AI responses
• Reduced hallucinations

⚡ FastAPI Backend

• High-performance REST API
• Modular architecture
• Easy deployment
• Scalable design

🎨 Modern UI

• Clean responsive interface
• Interactive chatbot
• Prediction dashboard
• User-friendly experience

---

🏗️ Project Architecture

                User
                  │
        ┌─────────▼─────────┐
        │    Web Frontend   │
        └─────────┬─────────┘
                  │
          FastAPI Backend
         ┌────────┴────────┐
         │                 │
         ▼                 ▼
 ML Prediction        AI Chatbot
 (LightGBM)        (Gemini + RAG)
         │                 │
         ▼                 ▼
   Premium Result     ChromaDB
                      Knowledge Base

---

🛠 Tech Stack

Frontend

• HTML5
• CSS3
• JavaScript

Backend

• FastAPI
• Python

Machine Learning

• LightGBM
• Pandas
• NumPy
• Scikit-learn

Artificial Intelligence

• Google Gemini API
• Retrieval-Augmented Generation (RAG)
• ChromaDB

Deployment

• Docker
• Render

---

📂 Project Structure

Insurance-AI-Copilot/
│
├── backend/
├── frontend/
├── chatbot/
├── chroma_db/
├── models/
├── prompts/
├── utils/
├── static/
├── templates/
├── requirements.txt
├── Dockerfile
├── render.yaml
└── README.md

---

🚀 Installation

Clone Repository

git clone https://github.com/Felix066/Insurance-AI-Copilot.git

cd Insurance-AI-Copilot

Install Dependencies

pip install -r requirements.txt

Configure Environment Variables

Create a .env file

GOOGLE_API_KEY=YOUR_API_KEY

Run the Application

uvicorn main:app --reload

Open

http://localhost:8000

---

💡 How It Works

Premium Prediction

1. User enters personal information.
2. Data is preprocessed.
3. LightGBM predicts insurance premium.
4. Prediction is displayed instantly.

---

AI Chatbot

1. User asks an insurance question.
2. Relevant documents are retrieved from ChromaDB.
3. Retrieved context is passed to Gemini.
4. Gemini generates an accurate answer.

---

📊 Machine Learning

Model Used:

• LightGBM Regressor

Pipeline:

• Data Cleaning
• Feature Engineering
• Encoding
• Model Training
• Prediction

---

🧠 AI Pipeline

Question
      │
      ▼
Embedding
      │
      ▼
Vector Search
      │
      ▼
Relevant Documents
      │
      ▼
Gemini LLM
      │
      ▼
AI Response

---

🔮 Future Improvements

• User Authentication
• Policy PDF Upload
• Conversation Memory
• Admin Dashboard
• Premium History
• Analytics Dashboard
• Voice Assistant
• Multi-language Support
• Cloud Database Integration
• Explainable AI (XAI)

---

🎯 Applications

• Insurance Companies
• Policy Recommendation Systems
• Customer Support
• Insurance Education
• AI Assistants
• Digital Insurance Platforms

---

🤝 Contributing

Contributions are welcome.

1. Fork the repository
2. Create a new branch
3. Commit changes
4. Open a Pull Request

---

📄 License

This project is licensed under the MIT License.

---

👨‍💻 Author

Felix Danie Jose

B.Tech Computer Science (Artificial Intelligence)

GitHub: https://github.com/Felix066

LinkedIn: www.linkedin.com/in/felix-danie-jose

---

⭐ If you found this project useful, consider giving it a Star!

It motivates future development and helps others discover the project.