Every upload, identified.
Each file you add is matched to its published record, and its details say how: by DOI, arXiv id or title.
open papers for deep search
Deep search finds open-access papers in OpenAlex, screens them against your question, and with your OK brings them in and answers from them. Every upload is matched to its published record, too.
step into the library Ask with deep search on, or set a watch on a topic.Live from OpenAlex: the same search deep search runs, open-access papers with a PDF copy only.
From one small action to what you see,
each step opened in the code.
Up to three searches, open-access papers with a PDF copy only: OpenAlex fetched 10 of 10 of those, where publisher links managed 2 of 20.
Abstracts are screened against your question in Markovian steps; at most eight sources reach you, best first.
The PDF comes from OpenAlex's own copy, and is checked to be a PDF, not a login page.
Authors, venue, year and cited-by, found by DOI, then arXiv id, then a title close enough to be the same paper.
Specific jobs in the app.
Each one has a place you can point to.
Each file you add is matched to its published record, and its details say how: by DOI, arXiv id or title.
Watch a topic and new open-access papers land in your Inbox, five at most each run.
For a DOI, arXiv first, then every open copy OpenAlex knows with a PDF link. No free copy means Konkyo says so instead of saving an abstract page.
Before an article shows a paper's figure, OpenAlex says whether its licence is open.
The choices behind the behaviour.
Open whichever part you're curious about.
The API key goes in a header, never the URL, because URLs end up in logs. Each call has an 8-second timeout.
A watch reads OpenAlex through its Zetaris REST table; if that fails, the same search runs directly.
A watch's search matches the topic in title and abstract, newest first, open access with a PDF copy, 25 at a time.
A deep search downloads papers at $0.01 each, so guests get 3 a day and accounts 20.