For decades, software was built on one premise: one-size-fits-all. You designed a screen, a flow, a visual hierarchy, and everyone — from the CFO to the intern — swallowed the same one. Personalisation, where it existed, was cosmetic: a «Hi, [Name]», a recommendation based on your history, a dark mode if you were lucky. The skeleton never moved.

What changes now is structural. With generative models, the interface stops being something you design once and becomes something you generate every time. The layout, the typography, the tone of voice, the amount of data on screen, the imagery and textures themselves — all of it becomes an output of the model, calculated according to who’s on the other side. Software stops being a grey, gridded tool and turns into a kind of chameleon that adapts to each person’s visual psychology.