Building Qube's design team from zero

The problem

When I joined Qube in 2016 there was no design team. More than ten products served studios, distributors and theatres, and each one looked and behaved differently.

The company needed design to keep pace with engineering without slowing anyone down.

What I found

A small team cannot cover ten products by working harder. It needs shared foundations, and people placed where their strengths count.

Designers grow fastest by exploring on their own. Exploration without a deadline becomes wandering.

Key decisions

Place people by strength

Every person has a different superpower. On Qube XP, one designer defined the visual style; a later one carried execution and refinement.

Timebox exploration

I let designers explore on their own, then set a deadline matched to the size of the problem.

Hold a detail bar

Not perfectionism. Reviews focused on the details users notice, until the team spotted them without me.

Hire for attitude

Tools can be taught. Curiosity and work ethic can't. I also gave second chances with clear direction, and people almost always delivered.

Give the team one language

The Pixel design system let five designers and many engineers work from the same components, so a small team could cover every product.

Sit with engineering

Designers joined engineering conversations, not just handoffs. That is how an inbox redesign became a data-model change.

How it's built

Pixel design system

50+ components and tokens on BlueprintJS, documented in five tabs per component.

AI in the workflow

AI prototyping tools brought into the team's process to shorten design cycles. Late in my tenure, the design system documentation became searchable through NotebookLM.

AI matching engine

Designed an AI-driven engine that matches theatre records from many sources to one unified theatre database.