Running an OpenAI’s model on your internal Confluence documentation
Editorial note (01/13/2023): The article was corrected by replacing the term “ChatGPT” with “Codex” or “OpenAI’s model”. I am learning and I did not realize at the time that I was not using ChatGPT but a descendant of its engine, called Codex. I was calling a dog, a cat, but the story does not change and I hope you will find the material still useful.
Editorial note (05/16/2023): I am happy to see this article is drawing a lot of interest so with great power comes great responsibilities 😊. Per the suggestion of a reader, I created a GitHub repo to clone the code: https://github.com/fascani/QA-confluence-doc. I already made a couple of corrections. Also, I would replace the (hyper) basic Flask app with a Streamlit app like I explain here: https://medium.com/@francois.ascani/how-i-built-a-ask-me-anything-chatbot-web-app-with-gpt-3-python-and-streamlit-9f7e6ee4b4e1. Good luck!
Like many folks, I got excited about ChatGPT and read several articles about how it can change how we work. One question that popped out quickly in my brain was: Could I use ChatGPT on our internal documentation?
