Designing Adaptive AI Interfaces for Stella

Before AI became everyday, Stella explored what it meant for design to collaborate with intelligence. The goal wasn’t to launch a product but to understand how people might one day talk to data. We set out to create an interface that could learn, respond and adapt, long before those tools existed.

Machine learning was still early, training data was inconsistent and model output often unpredictable. Stella had to visualise intelligence that didn’t yet behave predictably. We built forward from assumptions, testing ways to make opaque systems feel understandable and trustworthy. Every decision balanced imagination with realism.

The work began with behaviour not pixels. We mapped how users might ask questions, how the system might interpret them and how it could respond visually. Early prototypes mimicked conversation through adaptive layouts and contextual feedback. I led the design direction, framing the product principles and working closely with data scientists to turn abstract models into tangible moments of interaction. Each cycle pushed the boundary between static UI and dynamic response.

The design treated AI as a collaborator, not a feature. Information surfaced progressively, motion reflected system thinking and errors were framed as learning moments. The interface spoke a universal visual language that balanced logic with empathy, a foundation that influenced later conversational systems.

Stella was never about perfection. It was about direction. The project proved that design can lead when technology is still forming, translating abstract potential into experiences people can actually imagine.

I’m Öncel, a multidisciplinary designer who creates products, systems and experiences people rely on every day. I’ve partnered with global brands and startups at different stages, taking ideas from zero to MVP, leading complex designs that reached millions of users and building foundational design systems that grow with teams.