After seeing the potential of generative UI using image models, I was curious about what other applications this could work for. Along with a good friend Tom Mackisack, we brought the idea to life using a Raspberry Pi and a 3D printer.
We explored how voice-first AI might look as an always-on companion, without traditional 'wake' words — which make existing products feel soulless.
Right now, existing AI experiences feel very one-sided. What if they could message first? We designed proactive suggestions — like finding cheap flights you mentioned wanting to book, noticing you only slept 5 hours, or suggesting a walk while it's still nice out.
We also explored how Orbal could act less like an object, and more like a friend. What if it could swivel to face you, or turn away to look at something? We designed emotional presence through subtle expressions and body language.
The round display became a canvas for AI-generated UI. While not amazing yet, this is the worst these image models will ever be.