An AI summarizer that reads the page you are already on

Lumen reads the document directly from the window you have open, so there is no upload step and no file to send anywhere. A research paper open in a PDF viewer works the same way as a specification open in a browser.

The summary is structured: the key findings, the definitions that matter, and the conclusions, with the fluff removed. You get the parts you would have highlighted after an hour of reading.

The same tool handles the things that are not quite PDFs. Long API documentation, a legal agreement, a pull request, a thread of email, a specification, a wall of release notes. Point Lumen at the region and it works out what the text says.

Text is extracted locally, so the summariser is reading the actual characters on your screen rather than guessing from pixels. That is what keeps names, numbers and version strings accurate.

A summary is a starting point, not the end of the conversation. Ask what a specific clause means, why the author made a particular choice, or how one section relates to another, and Lumen answers against the document it just read.

Free gives you 5 requests a day. Pro gives you 50, a stronger model for the harder documents, and Stealth Mode so the overlay stays out of your screen shares.