Ask HN: Has anyone successfully implemented a LLM for a knowledge base?

3 points by BOOSTERHIDROGEN ↗ HN

2 comments

[ 4.5 ms ] story [ 14.4 ms ] thread
I recently made a demo to query the contents of video recordings with LlamaIndex[0].

I then created a little app of my own with a backing vector store to let me query a specific type of creative content that I need to reference often in my writing. It's still a very rough WIP, but has been promising. It's nice to be able to conversationally query these things instead of trawling through hundreds of pages for some obscure bit of information.

Now thinking of making another store for all of my blog posts over the years. Sometimes I look up how to do something and stumble across one of my own posts when I Google it. And sometimes I _know_ I have a post about a certain topic and just need to find it for a quick reference.

[0] https://github.com/daily-demos/recording-vector-store