A generative interface uses context about the task to select or create components. An assistant might present a chart to explain a trend, then show a form when the user wants to edit a record.
An application can limit generation to approved components and rules. Well-defined inputs and interaction checks help keep the output consistent, accessible and usable.
ELI5
Imagine asking a weather assistant about tomorrow's rain and seeing a chart of the hourly forecast.
When you ask what to pack instead, the same assistant might show a checklist because the task has changed.
Does generative UI require writing new code each time?
The application can assemble existing approved components from a structured description without generating executable interface code.
Why constrain which elements the model can choose?
Constraints help keep interactions predictable and preserve design, accessibility and security rules.




