Three parts. One user model. Yours.

Readablog runs on a small, portable user model called WOTF. Three sections make up the whole thing.

Fingerprint: your privacy controls

Twenty-two bytes that describe your model’s settings, not you. Location with a precision dial (none, country, region, city, or precise). Confidence scores. Per-service flags that let you grant different services different permissions over the same model. A checksum that catches tampering.

Style Vector: how you want content delivered

Thirteen knobs from 0 to 255: intensity, verbosity, formality, technicality, abstraction, narrative, engagement, humor, and five media preferences (text, image, audio, video, links). This is how Readablog can serve you content shaped to how you read, not just what you read.

Embeddings: what you’re interested in

A list of interest strands. You start by writing a few sentences about what you like to read. Readablog turns that into a list of topics, each with a confidence value and a 256-dimensional vector that services can match against content.

Full walkthrough with visuals coming soon. For now, head back to the Home page or read about trust and privacy.