when working with Zotero, you need to finish reading pdf and then export annotations by Zotfile plugin to export pdf to a markdown file in obsidian, and the exported markdown file got refreshed every time when you want to re-export anything in that paper.
the point is that the reading experience in reference managers method is divided and not at smooth as Logseq style when you could copy the source in pdf and comment on that snippet any time you wish during the pdf reading and not have to experience any "stop to export“ process。