Search the Full Text of Every Paper You Have
Research Assistant reads the text of every PDF in your library, so a sentence you half remember is one search away. Your notes and highlights are searched along with it, and scanned papers can be made searchable without anything leaving your computer.
Every page of every paper
Global search finds a word or phrase in titles, authors, abstracts, your notes, your highlights and the full text of every PDF. Results are grouped by paper and show each match with the words around it.
Double-click a match and the PDF opens on its page, with every occurrence highlighted and arrows to step through them.
Narrowing down the library
The search box above the list matches titles, authors, journals and years, and understands filters typed into it:
tag:andcollection:for a tag or a collection.year:2020,year:2020-2024,year:>2020oryear:<2024for publication years.is:starred,is:unreadandis:readfor your own marks.is:recentfor the last ten papers you opened, newest first.
They combine with the collections and tags picked in the sidebar, where Starred, Unread, Read and Recent are one click away. Recent together with a collection brings back the papers from that project you were reading last. Every active filter appears as a chip you can remove, and Save keeps the whole set in the sidebar for next time.
Search by meaning
Switch on Semantic and the list is ranked by what papers are about rather than by the words you typed. The embedding model behind it runs on your computer, the same one that lets chat find the passages that answer a question.
Scanned papers
Pages that are only an image are detected while the text is read, the paper gets a scanned badge, and global search tells you how many papers it could not fully search. Make searchable recognizes the text offline, about a second or two per page, and adds it to the PDF as an invisible layer. The original file is kept, so the change can be undone on any of your devices.