We had interns this summer at Memgraph [1] that have built ODIN - Obsidian Driven Information Network [2].

They have written two blog posts "Obsidian Note-Taking with ODIN: Intern's Perspective" [3] and "Building a Backend for ODIN and RUNE: How to Make a Knowledge Extraction Engine" [4] on this topic. The cover the reasoning why the did some things in the way that they did, and they do talk about LLMs. Also do take a look at "RUNE — Our Journey to Creating a GitHub LLM Analytics Tool: Intern’s Perspective".

[1] https://memgraph.com/

[2] https://news.ycombinator.com/item?id=37597201

[3] https://memgraph.com/blog/addobsidian-note-taking-with-odin

[4] https://memgraph.com/blog/building-backend-odin-rune-knowled...

[5] https://memgraph.com/blog/rune-creating-github-llm-analytic-...

I stopped using Obsidian, etc for this reason. I spent a fair amount time creating a database but never got real value out of it. It was just a waste of time. I'm questioning the whole premise of these systems. Why do I need a personal database when I can find anything I need on demand online?

I now have a few simple text files with project ideas, book notes, and todos. It feels much better and I don't spend time managing it.

I've found the same thing. Systems like these feel like a one way street that consume time and information and don't give much in return.

IMO You will need to keep fooling around (rereading, linking, reclassifying and thinking about it all) for meaningful insights to arise. Personal insight can't be just "extracted", especially if you don't have a clear goal in mind. It will magically rise within you (and not on the computer screen) when you are fully immersed in all the data. So don't expect the database to do the work for you. It is just there to make it easier and give you a better foundation to work with.

LLm could help bridge that gap a little more, but never 100%.