📄 RAG-Powered Document Reader Chatbot with Gemini

This project is a smart chatbot that can read your PDF, DOCX, or TXT documents and answer any questions about their content.

It uses a Retrieval-Augmented Generation (RAG) approach with Google Gemini AI, meaning it first finds the most relevant text from your document and then asks Gemini to answer your question in a natural way.

Even a child can think of it like this:

🏫 Imagine you give your teacher a book (your document)
🙋 You ask the teacher a question about the book
🧠 The teacher looks at the book, finds the right part, and then explains the answer to you clearly

That’s exactly what this chatbot does!

✨ Features

✅ Upload PDF, DOCX, or TXT files
✅ Ask questions in normal language
✅ Uses smart AI (Gemini) to find answers
✅ Keeps a chat history so you can see all past questions and answers
✅ Clear chat & document anytime
✅ Download chat history as TXT

📦 Installation

1️⃣ Clone this project

git clone <your-repo-url>
cd RAG_Powered_Document_Reader

2️⃣ Create a virtual environment & activate it

python -m venv venv

# Windows
venv\Scripts\activate

# macOS/Linux
source venv/bin/activate

3️⃣ Install required dependencies

pip install streamlit PyPDF2 python-docx sentence-transformers google-generativeai

🚀 How to Run

1️⃣ Set your Gemini API key

In your terminal:

set GOOGLE_API_KEY=your_api_key_here  # Windows
export GOOGLE_API_KEY=your_api_key_here  # macOS/Linux

2️⃣ Run the Streamlit app

streamlit run rag_chatbot.py

3️⃣ Open your browser
Go to http://localhost:8501

🏗️ How It Works

Extract text → When you upload a document, the app extracts all the text.
Find best matches → When you ask a question, it finds the most relevant parts.
Ask Gemini → It sends the relevant text + your question to Google Gemini AI.
Show answer → Gemini gives a nice, simple answer.

📁 Project Structure

📂 RAG_Powered_Document_Reader
 ├── rag_chatbot.py        # Main Streamlit app
 ├── file_loader.py        # Extracts text from PDF/DOCX/TXT
 ├── qa_engine.py          # Handles embeddings + Gemini QA
 ├── requirements.txt      # Dependencies list
 ├── .gitignore            # Ignore unnecessary files
 └── README.md             # This file

🖱️ How to Use

1️⃣ Upload any PDF/DOCX/TXT document
2️⃣ Type your question in the text box
3️⃣ The chatbot will read your document & answer
4️⃣ You can clear chat + document anytime
5️⃣ You can download chat history as TXT

🎯 Example

You upload a Company Policy PDF
You ask:

“What is the company’s leave policy?”
The chatbot finds the right section and answers:

“The company allows 20 paid leaves per year and requires 2 days prior notice.”

🛠️ Tech Used

Streamlit → Web app interface
PyPDF2 / python-docx → Extract text from documents
SentenceTransformers → Find most relevant text chunks
Google Gemini AI → Generate natural answers

📜 License

This project is free to use for learning purposes.

Now you can ask your documents anything, just like chatting with a super-smart friend! 🚀

CoderRakeshSharma/RAG_Powered_Document_Reader