VidhAI: AI Legal Assistant

🧾 About This Application

BNS QnA Chatbot is an AI-powered legal assistant designed to help users understand the Bharatiya Nyaya Sanhita (BNS) — the updated Indian Penal Code. Built using Retrieval-Augmented Generation (RAG) and OpenAI's powerful language models, this chatbot fetches the most relevant sections from the latest BNS documents to answer user queries with high accuracy and context awareness.

🔍 Key Features:

• 📘 BNS-Aware: Trained on the latest Bharatiya Nyaya Sanhita files.
• 🤖 Conversational Interface: Ask questions in natural language.
• 📄 Contextual Retrieval: Uses RAG to fetch accurate, up-to-date legal references.
• ❗ Fact-First Answers: No hallucinations—if an answer isn’t in the context, it tells you.
• 🌐 Streamlit Frontend: Clean and responsive UI for a seamless chat experience.

Tech Stack

Frontend

• Streamlit: Web application framework

Backend

• FastAPI: Web server
• Langchain: MLOps & orchestration
• Pinecone: Vector Database

Models

• text-embedding-3-small: for embedding text into vectors
• gpt 4o-mini: LLM

Workflow

Repository Structure

bns-rag/
├── controllers/   # Application logic controllers
├── db/            # Database interaction components
├── frontend/      # Streamlit UI elements
├── model/         # OpenAI model setup
├── routes/        # API routes for various functions
├── temp/          # Temporary file storage
├── utils/         # Utility functions and helpers
├── .gitignore     # Git ignore file
├── main.py        # Main application entry point
└── requirements.txt # Required dependencies

Installation and Setup

Prerequisites

• Python 3.7 or higher
• pip (Python package manager)
• Git

Installation Steps

1. Clone the repository:

   git clone https://github.com/nakul-krishnakumar/bns-rag.git
   cd bns-rag

2. Create a virtual environment (recommended):

   python -m venv venv
   source venv/bin/activate  # On Windows: venv\Scripts\activate

3. Setup environment variables:

   OPENAI_API_KEY=your-openai-api-key
   PINECONE_API_KEY=your-pinecone-api-key
   PINECONE_INDEX_NAME=your-pinecone-index-name
   PINECONE_NAMESPACE=your-pinecone-namespace

4. Install dependencies:

   pip install -r requirements.txt

5. Launch the server:

   python main.py

6. Launch the application:

   streamlit run frontend/app.py

7. Open your browser and navigate to the URL displayed in the terminal (typically http://localhost:8501)

Contact and Support

For questions, suggestions, or issues, please contact:

• Developer: Nakul Krishnakumar
• Email: [email protected]
• GitHub Issues: Submit issues through the repository's issue tracker