Smart Study Buddy

A study assistant that answers questions about your lecture notes using RAG (Retrieval-Augmented Generation). You upload a PDF, it gets embedded into a vector database, and then you can ask questions about it through a chat interface.

What it does

• Searches through your documents to find relevant sections
• Uses GPT to generate answers based only on your materials
• Remembers the conversation so you can ask follow-up questions

Before you start

You'll need:

• Python 3.10 or higher
• An OpenAI API key
• A Pinecone API key (free tier works fine)

Setup

Clone the repo and cd into it:

git clone https://github.com/khalilCodeX/Smart-Study-Buddy-.git
cd Smart-Study-Buddy-

Create a virtual environment:

python -m venv venv
source venv/bin/activate

On Windows use venv\Scripts\activate instead.

Install the dependencies:

pip install langchain langchain-openai langchain-core langchain-community
pip install pinecone pypdf python-dotenv gradio

Create a .env file in the project root with your API keys:

OPEN_AI_KEY=your_openai_api_key_here
PINECONE_KEY=your_pinecone_api_key_here

Adding your own PDF

Put your PDF in the project folder and update the path in datastore.py:

self.file_path = "./your-document.pdf"

Running it

First time only - you need to embed your document into Pinecone. Open chat.py and uncomment the embed_documents() line, or run this in Python:

from chat import chat
chatbot = chat()
chatbot.embed_documents()

This splits your PDF into chunks and uploads them to Pinecone. It only needs to be done once per document.

Then start the web interface:

python gradio_chat_app.py

Open http://localhost:7860 in your browser and start asking questions.

Project structure

chat.py              - main chat logic
datastore.py         - loads and splits the PDF
vectorstore.py       - handles Pinecone operations
management.py        - initializes the API clients
gradio_chat_app.py   - the web interface
system_prompt.md     - instructions for the LLM

Common issues

If you get an "index not found" error, you probably haven't run embed_documents() yet.

If API keys aren't working, double check your .env file uses OPEN_AI_KEY and PINECONE_KEY exactly.

If you're getting empty responses, check the Pinecone dashboard to make sure your vectors were actually uploaded.

Customization

You can change how the AI responds by editing system_prompt.md.

To adjust how the PDF gets split into chunks, look at the RecursiveCharacterTextSplitter settings in datastore.py. The default is 1000 characters with 200 character overlap.