Analyzed before you read
When a book is added, a model reads it in overlapping windows and scores every passage for cognitive density and lexical complexity. Nothing runs while you read; the pacing is baked in ahead of time.
LLM-paced speedreading
You have more worth reading than hours to read it. beetread shows one word at a time and holds each for exactly as long as it deserves: longer for a dense idea or a hard word, with a breath at each clause. You move fast and keep it.
For twenty years, you survived online by skimming: scan, triage, move on. Then language models got good at summaries. The scarce human skill now is taking something difficult all the way through and keeping it. That's the reading worth speeding up.
When a book is added, a model reads it in overlapping windows and scores every passage for cognitive density and lexical complexity. Nothing runs while you read; the pacing is baked in ahead of time.
Each word is shown for a duration scaled to how much work it asks of you. A common word flashes by; a dense, abstract one lingers so your comprehension keeps pace with your eyes.
At clause and sentence boundaries the stream takes a short breath-pause, the way a good reader-aloud does, giving each idea a moment to land before the next one arrives.
The reader is open now with a small library of public-domain classics, the kind of books that reward going all the way through. Uploads are next: the paper you keep meaning to finish, the book on your nightstand. Join the beta and we'll email you the moment uploads open.