How should intelligence show up?

As part of our work to rethink consumer devices, we explored how people might interact with intelligence built into the operating system. A chat window felt like only one of many possible interfaces. If you were already holding the phone, selecting something on screen might be the easiest way to interact. On a walk, speaking through your earbuds made more sense. On a watch, you might only want a quick glance or a small action.

The output could change in the same way. Sometimes all you needed was a vibration or a sound to know something had happened, while another request might need a sentence, a map, an image or something you could actually manipulate.

An interaction could also move between devices and modalities. You might ask for something through your earbuds while walking, then move to the phone if the response needed a map, an image or a richer interface.

I kept coming back to earbuds because AirPods had already made it normal to spend hours with a small computer in your ears. As AirPods Pro started taking on things like hearing assistance, it seemed increasingly plausible that people might wear them for much longer parts of the day. With access to context, memory and actions, earbuds could become another way into the same system rather than simply an accessory to the phone.

That also made us question how much visual identity the system needed at all. The first versions were quite expressive, with eyes, mouths, blinking and different reactions, but over time we stripped more of that away and moved toward increasingly abstract forms.

We explored blobs, swarms, inkblots and shapes moving around each other, looking for ways for the system to feel present without turning it into a fixed character or personality. I never felt the system needed one interface or one visual identity. It could be a voice in your earbuds, a few words on a watch, a small animation or a full interface on the phone, depending on the context and how much information the interaction actually needed.