Five papers,
every morning.

A daily arXiv summary page has to do two jobs at once. It has to be fast enough to read before a standup, and serious enough that a technical reader can tell whether a paper is worth opening. Perceptum is built around that tension. It ranks new papers, keeps the current daily top five and weekly top twenty-five fresh, and summarizes each selected paper at three levels of detail.

Percepti does the first pass like a tireless research assistant with an editor's taste. It looks at author signal, topic fit, early citation context, reference density, and practitioner buzz, then turns the selected papers into plain-English summaries. The rankings are meant to be useful, not final. They are a starting point for attention.

Beginner summaries explain the problem and why it matters. Intermediate summaries read like notes from a sharp colleague. Expert summaries spend more time on method, result, caveats, and what the paper builds on. You can skim the headline, read the one-line thesis, or settle into the full five-minute version.

The product is intentionally small. There is no endless archive to browse and no pressure to clear a queue. Papers age out after about a week, because the homepage is meant to feel like a fresh research desk, not a reference library. Shared paper links still work while the paper is fresh, but discovery is focused on the pages where readers can subscribe and return tomorrow.