A personalised arXiv, built for you
inveni is an AI- and ML-powered lab for working with scientific literature. Our main feature is personalised arXiv updates: we scan every preprint posted that day and build a focused selection for each user, based on their interests.
Right now we train profiles in physics, mathematics and computer science.
Less searching, nothing missed
inveni is built to be quick to check and easy to live with. Here is what it does for you.
Save 90% of your search time
A hundred-paper feed condensed to a handful of suggestions — each with a figure preview and an AI summary.
Never miss what matters
Stay on top of what others in your field are publishing, know what your competitors are doing, and stay at the research frontier.
We're just getting started
Smart paper search and an AI research assistant grounded in millions of published papers are on the way.
Up and running in a few minutes
To send you AI-curated suggestions, we first need to understand your interests. We'll ask for at least 10–20 papers that cover them — field classics or your own work. Pull them straight from your Zotero library or Google Scholar profile, or just search by title or author.
-
Add your papers
Bring in the work you care about. Import your publications from Google Scholar or your Zotero library, or add papers one at a time by title, author, link, or DOI.
-
Let it learn
inveni reads through what you added and works out the themes, topics and directions that matter to you — no questionnaires, no manual tags.
-
Open your daily feed
From the next day on, a focused selection of new papers is waiting for you. Rate them as you go, and the picks keep getting closer to what you want.
Tip: the papers you add are not just your CV — they tell the model what you'd want to read. Include around 20 or more, covering everything you follow, not only your own work.
Tell us when something is off
Hit a bug, a long wait, or something you don't like? Press the Report a bug button — it captures a screenshot and the timing, and your note tells us exactly what to fix.
Anything else on your mind? Get in touch at [email protected].