Searching my own notes meaning instead of keywords. I use a small model to do the search and a mid-sized model (around 7-8B) to answer, pointed at a few years of my notes. It's not as smart as a cloud model, but it's private. I though things at it which I would never paste into ChatGPT. And its quick enough that 'search what I have thought about before' is now a habit instead of hassle. The hard part is getting it to surface the right notes, not the answering itself.
There are dozens of things you can find a list of things you can do here https://www.kdnuggets.com/5-cool-things-i-did-with-local-lan....
I always loved experimenting, building HomeAssistant, and tbh everything you don't want any public AI to know about you
Searching my own notes meaning instead of keywords. I use a small model to do the search and a mid-sized model (around 7-8B) to answer, pointed at a few years of my notes. It's not as smart as a cloud model, but it's private. I though things at it which I would never paste into ChatGPT. And its quick enough that 'search what I have thought about before' is now a habit instead of hassle. The hard part is getting it to surface the right notes, not the answering itself.
I use qmd for that on over 20 years of journaling and saved quotes, works great
https://github.com/tobi/qmd
There are dozens of things you can find a list of things you can do here https://www.kdnuggets.com/5-cool-things-i-did-with-local-lan.... I always loved experimenting, building HomeAssistant, and tbh everything you don't want any public AI to know about you