The products I've designed are different. The way I approach them isn't.

People often think of design as the interface. But that's far downstream of what I think is one of design's most valuable outputs: shared understanding.

Good products don't emerge because one person has the best idea. They emerge when a group of people finally see the same problem in the same way.

That's why I spend so much time facilitating workshops, creating frameworks, sketching models and building playful exercises. They're not activities that happen around design - they're often the design work itself.

Being a language graduate and deeply interested in linguistics and semiotics, I give words careful consideration.

That's why one of the most powerful design tools I've discovered isn't visual, it's linguistic. Here's how it works:

Giving something a name doesn't simply make it easier to reference. It creates a shared mental model that people can interrogate, build on and ultimately use to make better decisions. Some examples of this practice can be viewed as critical moments in my case studies.

In discovery cases, a name is required to literally bring something novel to life. But I've found that even when something is fully known, teams can recognise a phenomenon long before they have a shared way of talking about it.

Shared uncertainty can be investigated. Private uncertainty tends to become politics.

This isn't branding. Or a fondness for memorable phrases. Language changes what groups are able to think about together.

Once something has a name, it stops being a vague feeling and becomes an object that can be questioned, challenged and improved.

AI doesn't replace my thinking. It multiplies the number of conversations I can have with my design.

Rather than treating AI as an answer engine, I increasingly use it as a thinking partner.

Sometimes it's a researcher. Sometimes a critic. Sometimes a prototyper. Sometimes an engineer. Sometimes I'll make it argue with me until I understand my own thinking more clearly.

That shift has changed my design process far more than any particular model or tool.

My favourite tool I built that I keep coming back to is my virtual design crit room, "populated" with my design heroes from differing disciplines (and principles).

I give them a problem or a mockup or a hypothesis and then watch them unpack it. It explicitly upends the question-answer relationship of most AIs and I had to carefully craft the instructions to avoid normal instincts to round off or conclude investigation.

Cursor has also fundamentally changed where my design process ends. Instead of stopping at mock-ups, I increasingly build working products - and while that's fun in and of itself, I do that because interactive software is a much better thinking medium than static pictures.

This can be evidenced in the creation of this portfolio:

Occasionally the design problem isn't the product; it's the absence of a good thinking tool. In those moments I'll happily spend an hour building a disposable tool if it lets me understand the problem more deeply.

On the homepage, I eventually needed motion-tracking debug and manual keyframe editing to ensure a smooth transition in the headline.