I Kill Features for a Living

I remember the exact moment I discovered Word Art. I was at my little village primary school in Cornwall, sitting in front of a chunky beige monitor, and someone showed me PowerPoint. Animations where text could fly in from the side, words that could bounce, and then the holy grail itself, Word Art, which let me make my name curve like a rainbow and shimmer with gradients and look like it had been carved from chrome. I spent entire afternoons making things that belonged on the side of a monster truck.

It was, obviously, hideous, and so was everyone else's. For a few glorious years, every school project and church newsletter in the country looked like it had been designed by someone having a fever dream about a rave. Word Art gave us all access to "design" without giving anyone taste, restraint, or any sense of when to stop.

I've been thinking about Word Art a lot lately, and it started with a conversation I had a few years ago with Greta Chambers, who's Head of Design Ops at Dentsu. We were chatting over lunch about where interfaces were heading, and we got into this idea of UI becoming fully personalised, generated on the fly, adapted to you, built for your specific needs in that specific moment. A fundamental shift in who gets to shape what software looks like. It felt speculative at the time, but watching how quickly things have moved since, I think we were just early, and my first thought seeing it all unfold was that we're about to live through Word Art all over again.

My dad trained as a graphic designer, and he has this story that puts all of this in perspective. Early in his career, if he wanted to create a graphic, he had to code the vector coordinates by hand, literally writing out the numerical position of every point, then walk it to another building to compile, then wait for it to go to a printer. Days of work for something we'd now knock out in minutes.

That laborious process is basically how building UIs has worked for most of software history. If you want to change a button colour, that's a ticket, and a designer mocks it up, and an engineer implements it, and it goes through review, and it gets deployed, and maybe you see it in production in a week if you're lucky. The distance between having an idea and seeing it realised has always been enormous, and while it's never been as bad as walking between buildings with punch cards, it's the same fundamental constraint where idea and implementation are separated by process.

That gap is collapsing though, and I saw it firsthand when I met with Jake Wells recently, co-founder at Meshed. He's been building products for years and used to run teams as a Product Owner with a proper enterprise background, but now he uses Claude Code to make design changes while he's on user testing calls. Someone gives feedback, he makes the change live, continues the conversation, and thirty seconds later he's validating whether the fix actually works. It's fabulous to watch, and while it's still technical and Jake clearly knows what he's doing and this isn't drag-and-drop yet, it's a world away from my dad's building-to-building workflow. The feedback loop has shrunk from days to seconds, and it's only going to get easier until this kind of thing is available to people with no technical background at all, which is exactly what Greta and I were talking about and exactly what worries me about the Word Art problem.

Because when I was eleven, Word Art didn't make me a designer. It made me someone who could produce design-shaped objects without understanding why any of my choices were good or bad. I had access to the tools but none of the judgment, and the results were predictably chaotic.

MC Dean wrote about this recently in her piece "Design Just Got Serious Again," where she argues that AI hasn't made design easier but has instead made surface-level design universally accessible while raising the bar for genuinely good work. When anyone can prompt their way to a competent interface, process stops being the differentiator, and what matters is taste and judgment and deep craft, the stuff you can't shortcut. She draws a comparison with desktop publishing in the nineties, pointing out that QuarkXPress didn't kill graphic design but instead eliminated the production artist role, the part of the job that was mostly about operating tools. The designers who understood typography and composition and visual hierarchy became more valuable, not less, because the craft beneath the tooling was what survived.

I saw this play out firsthand growing up. There was a print shop in Lostwithiel called Palace Printers, and I used to spend hours there because I was friends with the owner's son, learning how everything worked. They had this old Heidelberg press, a proper mechanical beast, and later digital printing too, and when desktop publishing took off everyone assumed places like that would disappear because why would anyone pay for professional printing when you can do it at home? But Palace Printers stayed busy. The quick stuff moved in-house, sure, but the work that actually mattered, where craft and quality counted, still went to the people who knew what they were doing.

I think we're heading somewhere similar with interface design. The basic stuff will get absorbed into AI tooling, generated on the fly and personalised to the moment, and that's probably fine and maybe even good. But the complex, considered, user-centred work will still need people who understand why certain choices work, people with the judgment that eleven-year-old me definitely didn't have when I was making my name shimmer like chrome.

My dad's been a graphic designer for over forty years now, and when CAD first appeared people said his profession was finished and that anyone could do it. Forty years later, graphic design is still very much a thing because while the tools changed completely, the need for people who actually understand the craft didn't go anywhere.

So I'm not worried about designers disappearing. I'm more curious about what happens in the messy middle, the period where the tools are available to everyone but the taste hasn't caught up, the Word Art phase. We might be about to live through a few years where the average quality of interfaces actually gets worse even as the tools get better, and then eventually the people with real craft will matter more than ever, the palace printers of interface design.

I don't know exactly how it plays out, but I suspect we're about to find out.