adityasharmadotai-hash/docs-reader-rag-agent

This project is a small chatbot that reads your Word documents and answers questions about them. You drop one or more .docx files into a folder called docs/, run the app, and then type questions in a chat box. The app sends your documents and your question to OpenAI's GPT-3.5 model, and the answer comes back in the chat. The author positions it as a beginner-friendly introduction to Retrieval-Augmented Generation, the technique of looking things up in your own files before asking a language model to write an answer. The code is split into three short Python files. loader.py reads every .docx file in the docs/ folder, including both the regular paragraphs and the tables inside the document. agent.py packages those documents up with the user's question and sends them to OpenAI. app.py is the Streamlit web interface that shows the chat and the clear-chat button. Features called out in the README include multi-document support, reading both paragraphs and tables, a rule that the bot should only answer from the supplied documents rather than make things up, and a clarifying-question behaviour when several documents are loaded at once. To run it locally, you clone the repo, install the requirements with pip, drop .docx files into docs/, create a .env file with your OpenAI API key, and start the app with streamlit run app.py. The README also walks through deploying it for free on Streamlit Cloud by pushing to GitHub and pasting the API key into the Secrets section. The license is MIT.