Basic RAG

This is a simple agentic Retrieval-Augmented Generation (RAG) system that can be used as a template for the projects. It is built in LangChain and LangGraph with Chroma and OpenAI support.

Current Features

• Upload and manage documents in the local Chroma vectorstore
• Query an LLM based on uploaded documents
• Query an LLM to perform web search with Tavily (optional)

Prerequisites

Ensure you have installed:

• Python 3.13
• uv

Installation

1. Clone this repository

git clone https://github.com/a-rogalska/basic-rag.git
cd basic-rag

2. Add necessary environment variables to the .env file

   LangSmith and Tavily env variables are optional

3. Install packages

uv sync

4. Run streamlit app

#Windows
.\.venv\Scripts\activate
# execute this in the root of the project
streamlit run .\src\Welcome.py

Planned Features

• Voice support with local models
• Local LLMs support with Ollama