Gideon Bullock

I've spent most of my career in places where design had to prove it was worth having. Early-stage startups with no design function, large organisations where design was seen as a downstream service, environments where the work was figuring out how to make design matter before you could do the design itself.

I ran the Toyota function on a full P&L, and I have sat in investment discussions, governance boards and strategic planning as the person accountable for design's numbers as well as its quality. Design leaders who can't talk about resourcing, utilisation and margin in concrete terms don't get taken seriously at the level where decisions are made.

The part of the job I care most about is developing people. Setting a clear bar, then giving people the context and autonomy to meet it. Several of the team I built at Toyota have told me it was the best team they'd ever worked in. That matters more to me than most metrics.

When a product runs on a model, behaviour is the larger half of the design problem. What the system attempts. What it declines. How it responds to a signal it can't read. What it does once it has been running long enough to have a history. That behaviour is a surface in its own right and it needs designing with the same rigour as anything visible.

I designed how an empathic voice should respond in an interview practice session, then again for a clinical companion where the bounds are tighter. When PrepCall's quality started degrading by the fourth session, the cause was a system prompt that had grown six-fold on accumulated history. Hard character budgets now hold it inside a fixed envelope. Emotrix works a layer below that, on what a system is entitled to conclude from a signal before anything acts on it.

Few designers are doing this work yet. It's the part of the job I want next.