A solution for live recommendation would involve heavy computation, So it definitely cannot be local. Users who need live recommendation should be willing to upload their notes to a cloud service and then have good Internet connectivity to use the feature.
I personally feel these compromises are fair considering the major productivity gains from a live recommendation system, Do you feel that way?
I wanted to read an article in the economist associated with their podcast, I got some details from the podcast description(There were no links there). Here are the results from different search engines -
DDG, "economist deep mind 3 dimensional protein", First result is a 2020 article about how proteins fold [https://www.economist.com/science-and-technology/2020/11/30/how-do-proteins-fold], Related to my query but not the result I wanted.
Brave Search, "economist deep mind 3 dimensional protein", First result is a 2019 article about Deepmind and Google with a mention of protein folding [https://www.economist.com/1843/2019/03/01/deepmind-and-google-the-battle-to-control-artificial-intelligence], Not the result I wanted.
Google, "economist deep mind 3 dimensional protein", First result is from Aug 2022 titled "How artificial intelligence cracked biology’s biggest problem" [https://www.economist.com/alphafold-pod], Which was exactly what I wanted.
In this case Google got me the best result purely because I wanted the latest result, But the results from DDG might have better content and the podcast article links to it.
Before humans stored memories as zeroes and ones, we turned to digital devices of another kind — preserving knowledge on the surface of fingers and palms. Kensy Cooperrider leads us through a millennium of “hand mnemonics” and the variety of techniques practised by Buddhist monks, Latin linguists, and Renaissance musicians for remembering what might otherwise elude the mind.