[brucethemoose2]

First of all, this is a good idea and a cool app.

IDK if this is part of your plans, it seems like a perfect way to gather LLM training data... And a reasonable use case for LLM integration. And its not in explicit conflict with your privacy policy as far as I can tell, as long as you self host.

Everyone is training LLaMA on ChatGPT Q+A pairs, but a big corpus of filtered, purely human, human rated questions+responses would be so much better.