I still think it's possible to go incredibly far with a good data model, fuzzy search, automated tagging and feed publishing.
[1]: https://lgug2z.com/articles/the-bookmarking-data-model-is-wr...
[2]: https://lgug2z.com/articles/notado-07-2023-update/
[3]: https://lgug2z.com/articles/using-rust-chrome-and-nixos-to-t...
I would think about going one step further and having a full text search over my whole browsing history with a way to penalize or remove certain hosts. Then I would love to augment my web searches to also include things from my archive.
I feel I would have a much bigger utility out of this in comparison to thing through AI, though I get it could be also beneficial. However AI, I believe, is costlier to run. A nice, properly indexed, full text search probably could run on a phone at least for a private archive.
What specs do you think would be needed?