How I Built This Site, CMS and All, Without Knowing How to Code

Every blog has an awkward first post. Mine is awkward in a specific way: it is about the blog itself.

Three days ago this site did not exist. Not the design, not the address, not the place these words are stored. I am not a developer. I have never written a line of code, and I still have not written one. What I had was a clear idea of what I wanted and an AI assistant that could build it.

The video below shows this site on the day it was finished. The design is real. None of the writing in it is.

The travel stories, the hotel review, the photographs of places I have never been, the confident list of AI tools I supposedly use every day: all of it was demo content, written by Claude and loaded into the site with a single button so that I could see what a finished publication looks like instead of staring at an empty shell. The photographs were placeholders pulled from a random image service.

I kept it exactly long enough to film, then started replacing it with real work. Everything you read here now is mine: my own trips, my own photographs, my own opinions.

I am showing you the fake version anyway, because it is the only honest record of what this site looked like in the hour it was born. Think of it as the showroom model with the plastic still on the seats, photographed before anyone moved in.

I want to be precise, because that phrase gets used loosely.

I had never installed a development tool. I had never opened a terminal on purpose. I did not have an account on GitHub, and when it came time to upload the project I genuinely asked which of the two black windows on my computer I was supposed to be typing into.

That is the starting point. Not a shortcut, not a hidden technical background. Someone who knows what a good website should feel like and has no idea how one is made.

This is the part nobody puts in the exciting version of the story, and it is the part that mattered most.

Before I opened anything, I sat down and wrote a specification. Not code. A long, patient, slightly boring document in plain language describing the thing I wanted to exist.

I wrote down the things I did not want as clearly as the things I did. No heavy page builder. No monthly subscription to a platform that can change its rules.

I pasted the whole specification into Claude Code, which is the version of Claude that can read and write files on your own computer rather than only talk to you in a chat window. The model was Claude Fable 5.1. I pressed enter and went to make tea.

The first thing it did was check my machine, discover that I had neither of the two tools it needed, and install them. That alone would have stopped me for an evening on my own.

What arrived was not a template with my name dropped into it. It was an application, written from scratch for my specification, in about a hundred and twenty files.

Then it tested its own work. It started the site, opened it in a browser, took screenshots, noticed that images were not filling their frames correctly, fixed that, checked the layout at phone width, uploaded a test photograph through the real pipeline to confirm the resizing worked, and deleted the test afterwards.

I watched this happen. I did not intervene. I could not have intervened usefully.

I made an account on GitHub, which is where the instructions for building a website live, and created an empty project. Two minutes and one form.

The assistant did the rest. I asked my slightly embarrassing question about which black window I was supposed to use, and the answer was that I did not need one at all.

Worth knowing: my articles and photographs are not stored there. Only the instructions for building the site. The writing lives somewhere else, which is why publishing a new post is instant and involves none of this.

The site runs on Cloudflare. I made an account, and from there the assistant created everything: a database for the articles, a storage bucket for the photographs, and the site itself. A few minutes later there was a working address I could open on my phone.

One thing needed my card. The photo storage asks for a payment method before it can be switched on, even though the free allowance covers thousands of photographs and I have not been charged anything. I also chose my own admin password, which is the one thing I deliberately did not delegate.

The free address worked, but it was ugly and full of technical words. I wanted something I would not be embarrassed to say out loud.

I asked whether hubcurious.com was free. It was. I bought it, entered my contact details, and the assistant attached it to the site, arranged the security certificate, pointed the www version at the main one and switched the old technical address off.

**Ten dollars and forty-six cents a year.*" That is the entire running cost of this website.

There is no monthly platform fee waiting for me. Nothing here can raise its price, retire a feature I depend on, or decide that my kind of site no longer fits its terms. That was one of my requirements and I did not seriously expect to get it.

This matters, because "AI built my website" is the kind of sentence that quietly removes the person from the story.

None of that is programming. All of it is judgement, and none of it could have been skipped.

Three things.

The first is that the admin area is genuinely pleasant. I expected something raw and developer-shaped. Instead it is a clean editor where I write a heading, drop in photographs, paste a YouTube link and press Publish.

The second is that it explained itself. Everything it did was written down in plain language, in files I can reopen in six months when I have forgotten how any of it works. It answered every naive question I had without once making me feel that I should have known better.

The third is the one I keep thinking about. The hard part was never the technology. The hard part was knowing what I wanted. Once that existed as a clear document, the distance to a finished, working, genuinely good website was measured in hours.

Not to sell you anything. I have nothing to sell yet. :)

I am writing it because I spent years assuming that a site like this one was out of reach. Too technical, too expensive, too dependent on finding and paying the right person, too likely to end up abandoned half-finished. That assumption quietly cost me a lot of time.

It is not true any more, and that is the whole message.

AI did not replace a developer here so much as it removed the reason I never started. The taste, the judgement and the stubbornness still have to be yours. But the gap between I wish I had a site like that and I have a site like that has become astonishingly small for anyone willing to sit down and think clearly about what they actually want.

If you have been carrying an idea around, waiting until you know enough to begin: you do not need to know enough. You need to describe it well.

From here on, everything on this site is real: my own trips, my own photographs, my own opinions.

If you are reading this a year from now, this post will look like a strange little artefact from the week the site was born. That is exactly why I am keeping it.