RagE (Retrieval Augmented Generation Engine) is an intelligent system designed specifically for the insurance industry. It leverages FastAPI and retrieval-augmented generation to help insurance professionals and customers quickly find, understand, and analyze insurance policies, claims procedures, and coverage details.
- Policy Document Analysis: Upload and analyze insurance policy documents
- Natural Language Queries: Ask questions about policies in plain English
- Coverage Comparison: Compare different policies and their coverage details
- Claims Assistance: Get guidance on claims procedures and requirements
- Regulatory Compliance: Stay updated with insurance regulations and compliance requirements
- FastAPI Backend: High-performance API framework
- RAG Pipeline: Enhanced document retrieval and analysis
- Vector Database: Efficient semantic search for insurance-specific content
- Multi-Database Support: Works with PostgreSQL, MySQL, SQLite
- Responsive UI: Clean, modern interface for all devices
Insurance documents are dense, complex, and filled with specialized terminology. Traditional search approaches often fail to capture the nuanced relationships between insurance concepts. Our RAG system:
- Understands insurance-specific terminology and concepts
- Provides accurate answers by retrieving relevant policy sections
- Reduces time spent searching through lengthy policy documents
- Ensures responses are grounded in actual policy language
- Helps both insurance professionals and customers understand complex coverage details
- Python 3.8 or higher
- pip (Python package installer)
- Docker (optional, for containerization)
-
Clone the repository:
git clone https://github.com/yourusername/RagE.git cd RagE -
Install the required dependencies:
pip install -r requirements.txt -
Set up environment variables by copying the example:
cp .env.example .env -
Run the application:
uvicorn app.main:app --reload -
In a separate terminal, run the Pathway pipeline:
python pathway_pipeline.py -
Open your browser and navigate to:
http://localhost:8000