Policy ChatBot 📚

Overview

This project implements an AI-powered chatbot that answers questions based on company policy documents using Retrieval-Augmented Generation (RAG). The chatbot uses a company policy PDF file as its knowledge base and provides accurate, context-aware answers that are strictly grounded in the policy content.

Unlike general-purpose AI chatbots, this system ensures all responses are based solely on the provided policy documents, making it reliable for company-specific queries. It combines modern AI technologies like OpenAI's language models with vector embeddings and semantic search to deliver relevant and accurate information.

Demo

Interactive chat interface showing a query about company policies with source citations

Key Features

Interactive Web Interface - Clean, user-friendly chat UI built with Streamlit
Source Citations - Every answer includes references to the specific policy documents
Conversation History - Maintains context throughout the chat session
Semantic Search - Uses AI embeddings to find the most relevant policy sections
Grounded Responses - Answers are based only on the policy documents, reducing hallucinations

How It Works

1. Document Processing: Policy PDF files are loaded and split into manageable chunks
2. Embedding Generation: Each chunk is converted into vector embeddings using OpenAI
3. Vector Storage: Embeddings are stored in a Chroma vector database for efficient retrieval
4. Query Processing: User questions are converted to embeddings and matched with relevant document chunks
5. Answer Generation: The most relevant chunks are sent to GPT-5 to generate accurate, contextual answers

Setup

1. Install dependencies:

pip install -r requirements.txt

2. Create a .env file with your OpenAI API key:

OPENAI_API_KEY=your_api_key_here

3. Add your policy PDF files to the data/ folder

Usage

Step 1: Create the Vector Database

Run this first to process your policy documents:

python create_database.py

This will:

• Load all PDF files from the data/ folder
• Split them into chunks
• Create embeddings using OpenAI
• Store them in a Chroma vector database

Step 2: Run the Web UI

Launch the chatbot interface:

streamlit run app.py

This will open a web browser with an interactive chat interface where you can ask questions about your policies.

Features

Interactive Chat Interface - Clean, modern UI for asking questions Source Citations - See which documents the answers come from Chat History - Keep track of your conversation Semantic Search - Finds relevant information using AI embeddings

Technologies Used

• LangChain - For RAG pipeline
• OpenAI - For embeddings and chat completion
• Chroma - Vector database
• Streamlit - Web UI framework