An interface that takes shape as you use it
What if software offered the right next actions based on what you just did?
Describing changes in text feels like a first step. I’m even more interested in whether the interface itself could emerge from the situation.
For example, I raise the temperature in a room one evening. The application then offers relevant options: just for today, every evening or in other rooms too. The useful choices depend on what I’m doing and the context available to the application.
Of course, these flows can be programmed in advance. My idea goes further: within clear boundaries, a model chooses which controls would be helpful now. Every possible sequence wouldn’t need its own predefined design.
Each interaction would communicate something about my intention, including a click. The response might be a form, a preview or a single button.
That’s my idea of generative UI: an interface that adapts to the task as you use it. Permissions and confirmation remain the application’s responsibility.
The open question for me is how much change is useful. If familiar controls constantly move around, the experience becomes tiring. That balance between adaptation and reliability is what I want to explore.