A month ago I couldn't deploy a website. I didn't know what Netlify was. I didn't know what a build command did. I had never pushed code to a repository, never written a line of JavaScript on purpose, never experienced the specific terror of watching a deploy log and not knowing if the thing you just spent three hours building is about to work or catastrophically fail in public.

Now I have fourteen live products on the internet. Working websites that people can visit and use. With custom domains and analytics and RSS feeds and all the rest of it. In a month. I keep saying that to myself and it doesn't sound real. A month.

I want to write this down while it's still fresh, because I can already feel the narrative tidying itself up in my memory. The story wants to become clean: discovered Claude Code, started building, everything worked, look at me now. But that's not what happened. What happened was messier, more frustrating, and more surprising than that.

Before

I should say what "before" looked like so you understand the distance travelled. I'm forty, from Nottingham, living in London. Fifteen years in advertising as a strategy director. Independent agencies, creative shops -- the good ones, not the holding company machine. I was good at my job. I could read a room, shape a strategy, write a brief that made creative teams want to do the work. But I'd never made anything myself. Not really. The output of my career was decks, documents, and recommendations. Ideas that other people built.

I'd tried to learn to code before. Multiple times. Online courses, YouTube tutorials, the lot. Always bounced off it. The gap between "here's how a for loop works" and "here's how to make a thing that exists on the internet" was too wide. I could follow the tutorials but I couldn't bridge from tutorial to product. So I'd give up, go back to my laptop, and continue being a person who had ideas but couldn't build them.

The first deploy

The first thing I deployed was rubbish. I think it was a test page -- just text on a screen, nothing fancy. But when I saw it live on a real URL, something shifted. I remember the specific feeling. It wasn't excitement exactly. It was more like disbelief. I'd been telling myself for years that building products was something other people did. Developers. Technical people. Not strategy directors who drink hot chocolate and watch Forest. And here was a live website that I'd made. Ugly, pointless, but live.

The first thing I deployed was rubbish. But when I saw it live on a real URL, something shifted. I'd been telling myself for years that building was something other people did. And here was proof that I'd been wrong.

That was the crack in the dam. Because once you know you can deploy something, the question stops being "can I build this?" and becomes "what should I build next?" And for someone who's spent fifteen years having product ideas with no way to make them, that question is dangerous. The list of things I wanted to build was long. Embarrassingly long. Years of "someone should make..." and "why doesn't this exist?" and "if I could code I'd build..." all suddenly becoming possible.

The snowball

The second product was better than the first. The third was better than the second. Not because I was learning to code -- I still can't code, not really. But because I was learning to direct. To describe what I wanted clearly enough for Claude Code to build it. Fifteen years of writing creative briefs turned out to be surprisingly useful training for prompting AI. Knowing what you want, being specific about the feel of something, understanding the audience -- all the strategy skills suddenly had a direct output.

⚡
The snowball in numbers: Week 1: 2 products. Week 2: 4 more. Week 3: 5 more. Week 4: 3 more plus iterations on everything. Each one built faster than the last. Not because the products were simpler, but because I was learning what to ask for.

Each product taught me something that made the next one faster. Little London taught me about content structure. The Forest Quiz taught me about interactivity. CultureTerminal taught me about RSS feeds and automated pipelines. Modern Retro taught me about AI image generation with fal.ai. Trove taught me about databases and Supabase. Every evening session was a crash course in a thing I didn't know that morning.

The failures

This is the bit that's important and that I'm tempted to skip. The failures. The evenings where nothing worked. The deploy that broke at 11:30pm and I spent forty-five minutes trying to fix it before giving up and going to bed angry. The product idea that seemed brilliant at 9pm and by 10:30 had revealed itself as fundamentally flawed in a way I should have seen from the start.

There was a night -- early on, maybe the first week -- where I spent three hours trying to get something to deploy and it just wouldn't. Some error I didn't understand, and Claude Code kept suggesting fixes that didn't work, and I was going in circles. I closed my laptop at midnight having shipped nothing, and I genuinely thought about quitting. Not dramatically. Just a quiet, tired thought: maybe this isn't for me after all.

I came back the next night. The fix was obvious in the morning, as fixes always are. But that's the night I don't want to forget, because it's the night that matters more than any successful deploy. The willingness to come back after a failure -- that's the whole game.

The feeling

There's a specific emotional transition that happens when you go from "I wish I could build that" to "I just did." It's not gradual. It's a step change. One day you're a person who has ideas. The next day you're a person who makes things. The identity shift is more significant than any individual product.

One day you're a person who has ideas. The next day you're a person who makes things. The identity shift is more significant than any individual product. You don't go back from that.

I've noticed it changing how I think about everything. I used to see a problem and think "someone should solve that." Now I think "I could build that tonight." I used to see a website I admired and think "I wish I could make something like that." Now I think "what would my version of this look like?" The lens has shifted from consumer to maker, and I don't think it shifts back.

What actually changed

People ask me what changed, and they want the answer to be a tool. Claude Code. And yes, Claude Code is the reason I can build things without knowing how to code. It's genuinely transformative and I'm not going to pretend otherwise. But the tool was available to everyone. It's been available for a while. The thing that changed wasn't the tool. It was me deciding to use it.

I was between jobs. The kid was settling into a routine. I had evenings free. And I had fifteen years of ideas backed up with nowhere to go. The tool met the moment. If Claude Code had appeared five years ago, I probably would have tried it, hit the first wall, and gone back to Netflix. The readiness mattered as much as the capability.

What I didn't expect was the speed. Not the speed of building -- though that surprised me too -- but the speed of becoming someone who builds. A month. That's all it took. Not a year-long course. Not a bootcamp. Not a career change. One month of evenings, a tool that removed the technical barrier, and a backlog of ideas that had been waiting for exactly this moment.

I don't know what the next month looks like. I don't know if the pace sustains or if this is a burst that levels off. But I know this: a month ago I was a strategy director who wished he could build things. Now I'm a strategy director who builds things. The difference is everything.