Case Study · In progress

The design system that couldn't grow

At Viamagus I'm lead designer on Synthia, a middleware AI platform that automates workflows and optimizes processes at the organizational level. It's pre-MVP and under wraps — but the most interesting problem there isn't confidential, because it wasn't a screen. It was a diagnosis.

The diagnosis

Synthia had a design system already. It worked — for the product that existed. The components had been built around specific screens and flows, which meant every new feature either broke them or spawned a one-off variant. That's the most common way a design system fails, and it's rarely noticed as a failure: the system looks fine, the team just gets slower, and the product quietly fragments. It was built bottom-up, extracted from what had been designed already. It worked for the past and fought the future.

What I did

I rebuilt the components to compose rather than to match — designed against the shapes of problems rather than the shapes of existing screens — and modernized the visual language while I was in there, so the system reads as contemporary and cohesive rather than accumulated.

The part that was actually hard

It wasn't technical. To leadership, the work looked like a restyle — because the visible change was cosmetic and the load-bearing change was invisible. Making the case that this same work was a scalability fix — that consistency isn't decoration, it's what lets a small team ship fast without the product fragmenting — was the argument that mattered. Winning it in an engineering-led org, upward, without being the person who decides, is the work I'm proudest of there.

"They saw a restyle. What it actually was: the difference between a system built for the screens we had and one built for the features we hadn't designed yet."

Why this pairs with Zendr

Two design systems, opposite problems. Zendr: build one from zero, structured so it outlives the designer. Synthia: inherit one that's failing, find out why, rebuild it to scale — and win the argument for it. Greenfield and brownfield.