How I think about design systems

I don't design screens. I design the infrastructure that keeps decisions consistent long after the screens change.

The problem I keep solving

Across public health platforms, insurance systems, and manufacturing tools, I've seen the same failure repeat: design decisions get made once, then forgotten. A color changes without anyone remembering why the original one was chosen. A component gets rebuilt three times because nobody knew it already existed. The system doesn't fail because people are careless — it fails because nothing preserves the reasoning behind what was built.

What I actually optimize for

My work sits at the intersection of design, development, and operations — what's often called DesignOps. Three principles guide it:

Decisions need memory.

A design system isn't a component library; it's a record of trade-offs. If you can't explain why a decision was made, you can't safely change it later.

Workflows are part of the design.

The handoff between design and development is where most systems quietly break down. I treat that pipeline — not just the interface — as something to be designed deliberately.

Durability beats novelty.

Tools and frameworks turn over every few years. A system built to depend on one tool is a system with an expiration date. I build for standards that outlast the tooling around them.

How this shows up in practice

At RAMQ, this means building UX foundations for an organization before building screens. At Akinox, it meant designing the pipeline between design and development, not just the interface on top of it. In Agentica, my personal system, it means every meaningful decision is logged, versioned, and traceable — proof that this approach scales beyond a single project.

Where AI fits

AI agents can accelerate parts of this work — detecting inconsistencies, drafting documentation, proposing components. What they can't do is decide. Governance stays human, by design. My approach to AI in design systems isn't about automation for its own sake; it's about making sure automation never outruns accountability.