Blog
The IO Blog
Research Note
·The Whiteness Test: Benchmarking Five Models Across Twelve Reading Lenses
One paragraph of Melville, twelve production personas, five language models — 60 live generations, read line by line. Round 2 adds Google's brand-new Gemini 3.6 Flash and 3.5 Flash-Lite, a scripted compliance audit that caught an error in our own first edition, and a pricing ladder where the cheapest good model is still the one from January.
Read →
Reading Culture
·The Age of Reading Is Over. But Is It?
The Atlantic's new cover says it is. The decline is real — I've spent a year staring at the numbers while building against them. On where the blame actually belongs, and what an AI that refuses to summarize is for.
Read →
Reading Culture
·Why Read the Classics in the AI Era?
A machine can now compress any book into a tidy paragraph. The paragraph delivers the conclusion — and drops the only part that changes you.
Read →
Reading Science
·Why Vertical Reading Works
One paragraph at a time, full-screen, a format that clears away the visual noise competing with meaning, while keeping what speed-reading strips out — your freedom to look back.
Read →
Product
·Atomic Focus: One Sentence at a Time
Atomic Focus collapses the screen to a single sentence, which sounds austere until you feel what it returns to a crowded mind, the few slots of attention a hard line actually needs.
Read →
AI Features
·Meet Your Experts: IO’s Twelve AI Personas
Most AI reading tools just summarize. IO ships twelve personas instead, each a different way of staying inside a passage rather than skimming past it. Here’s the whole bench, and the science behind it.
Read →