For most of my career, I've been the person in the room with the idea. The strategist who writes the brief, shapes the narrative, and then hands it off to someone else to actually make the thing. That was the deal. You think, they build. It worked, mostly. But there was always this gap - a frustrating distance between seeing something clearly in your head and having it exist in the world.
Then, sometime in late 2025, something changed. AI coding tools got good enough that someone like me - a person who had never written a line of production code - could start building real products. Not prototypes. Not mockups. Actual, live, functioning things that people could use.
I've now built and shipped fourteen of them.
The gap that always existed
If you've ever worked in an agency, you know the feeling. You spend weeks refining a strategy, crafting a brief, articulating exactly what something should be and who it is for. Then it goes into production and what comes back is... close. But not quite right. Not because the people building it aren't talented - they are - but because something always gets lost in translation. The intent, the feel, the taste of the thing.
This is the gap I'm talking about. Not a skills gap. A translation gap. The distance between vision and execution when those two things live in different people's heads.
When I started using Claude Code as my building tool, that gap collapsed overnight. Suddenly the person with the vision and the person making the decisions in the code were the same person. Me. And that changed everything.
What I actually built
Let me be specific. In the space of a few months, I shipped fourteen distinct products:
Modern Retro started as an experiment - what if you could see today's brands through the lens of a 1970s retail store? It became an AI art project with its own visual language and aesthetic identity. Trove is a bookmarking tool that learns your taste over time, something I wished existed for years. CultureTerminal aggregates cultural business news the way I always wanted to read it - across advertising, design, fashion, and media in one place. EVERYWEAR does the same for wearable technology.
Curio is my attempt to rebuild something I loved from the early social web - surfacing the links that matter most by tracking what people are actually sharing. Little London was born from a very personal need: finding weekend activities for my son in London without wading through rubbish recommendations. London Pub Guide came from the same instinct but for a different audience. And Forest Quiz exists purely because I'm obsessed with Nottingham Forest and wanted to make something for fellow supporters.
Oishii London curates the best Japanese restaurants in London - born from a genuine obsession with Japanese food culture. Nihongo is a Japanese learning app with four levels from hiragana to restaurant Japanese, because I wanted to learn the language properly. First Out tells you exactly where to stand on the tube platform to be nearest your exit - the kind of micro-utility that saves minutes every day. Sociology of Capitalism scores 25 brands on the tension between cultural relevance and business performance, updated automatically twice daily. Taste OS turns brand taste into a scoring system - five dimensions, twenty points each, applied to twenty brands. And Forest Flash Cards is a Top Trumps-style card game featuring thirty Nottingham Forest players across three game modes.
Each one is live. Each one works. Each one was built by someone who, a year ago, couldn't have told you the difference between a function and a variable.
What surprised me
The technical part wasn't actually the hard part. That surprised me. The hard part - the part I turned out to be well-prepared for - was all the stuff that sits around the code. Deciding what to build and what to leave out. Understanding who the thing is for and what they actually need. Making it feel right, not just work right. Naming it. Positioning it. Giving it an identity.
Those are strategy skills. Advertising skills. The same muscles I had been training for fifteen years were suddenly applicable in a completely different context.
The other thing that surprised me was how much I enjoyed the craft of it. There's a particular satisfaction in obsessing over the spacing of a heading, the colour of a hover state, the exact moment an animation triggers. These are details that most people will never consciously notice. But they feel them. And coming from a world where I had always cared about those details but could never directly control them, having that power was intoxicating.
What this means
I'm not a special case. There are thousands of people in creative industries - strategists, designers, writers, marketers - who have spent their careers developing a deep understanding of audiences, aesthetics, and what makes things work. They have taste. They have ideas. What they've not had, until now, is the ability to build.
That is changing. Fast.
The interesting question isn't whether AI can write code. It can. The interesting question is what happens when the people with the best ideas and the deepest understanding of audiences suddenly have the ability to bring those ideas to life directly. Without permission. Without a dev team. Without a budget.
I think we're about to find out. And I think the results are going to be more interesting, more distinctive, and more human than anything the tech industry has produced by itself. Because the people entering this space aren't starting from the code and working backwards to find a problem. They're starting from genuine needs, real taste, and a lifetime of understanding what makes things resonate with actual people.
That isn't a small advantage. That is everything.
