wtf is rag-chatbot-template-starter?

This project is a starter template for building a chatbot that answers questions based on PDF documents you provide. The chatbot uses a method called Retrieval-Augmented Generation, which means it first searches your uploaded document for relevant passages, then uses that text to help generate an answer. Every answer includes citations showing the source filename, page number, and the specific text it used, so you can verify where the information came from. The backend is built with FastAPI and ChromaDB. It supports multiple language model providers, including OpenAI, Claude, Gemini, and a local option called Ollama. You can switch between them by changing a single setting in a configuration file. A notable feature is its retrieval-only mode, which works without any paid API keys. In this mode, the system returns matched passages from your documents directly, without calling a paid language model to write a summary or answer. Setup is straightforward. You can run it with Docker using a single command, or install it locally with Python. On first run, it downloads an embedding model that handles multiple languages, including English and Thai. The README provides four API endpoints: one to upload and index a PDF, one to list uploaded documents, one to delete a document, and one to ask a question. You interact with the chatbot by sending questions to the chat endpoint, and you can control how many source passages it retrieves per question. The README is upfront about what this free starter edition includes and what it lacks. The core pipeline, provider support, and Docker setup are all here. However, features like a built-in chat interface, streaming responses, multi-turn memory, OCR for scanned PDFs, and support for other file types are only available in a paid Pro version linked from the README. The project is licensed under MIT, allowing use in personal and commercial projects.