I've deployed it for my parents but in the end it's too mechanical in entering metadata, guessing rules works of course, but only for "regular stuff" and querie

I've deployed it for my parents but in the end it's too mechanical in entering metadata, guessing rules works of course, but only for "regular stuff" and queries are not that extraordinary, for me personally I've "solved" using org-attach meaning I use org-mode notes as metadata-rich bookmarks, fully searchable for my files and eventually rga if it's really needed (at the size of my docs it's fast enough and much simple than Recoll/solr), but for non-emacs-er I still have to find a flexible enough storage solution... I've even experimenting a generic usage of Zotero, does not work much, ideally these days we need to manage files NOT in a hierarchy but in a graph, automatically managed, annotating files with links in notes, being able to search titles, links, notes all together. Zim with attachments for non-teaches it's still limited, too tied to the underlying file system, Zotero and Paperless are way too mechanic and Paperless do not allow note, a separate Dokuwiki with links to Paperless stored docs it's simply way too much overhead... Long story short it's remarkable the automatic OCR (ocrmypdf), auto-classification, metadata automation etc but it's still not "the universal solution" IMO...