It took nineteen homepages to build this site
Most of what AI built for this site is in the bin. That's the whole point.
I built this site with Claude doing most of the typing. The cassette you’re looking at, the arcade cabinet/CRT tele on the lab page that I’m still not sure about, the way the type sits, all of it came out of a chat window (I’m using tabby on windows, not sure if I’m keen yet but if you’re not trying a new terminal every few days are you really vibe-coding?). And I know exactly how that reads in 2026. that must have been so difficult for you to type “make website” into an AI. how you have suffered for your art…
The bit nobody shows you, when they show off the AI thing they made, is the bin. Mine has a lot in it, I threw a lot of shit out.
Concept 06. Arcade cabinet centrefold. I genuinely liked this one for about an hour.
This site went through nineteen homepage concepts before I picked one. Some were genuinely nice. Most were not. There was a brutalist newsprint version, a CRT terminal, a thing that looked like a Bandcamp page, an arcade cabinet centrefold, a literal music zine, gradients that were very reminiscent of that one iPhone update. I killed them in batches. “I’m really not keen on 2 and 4.” “Discard 9 and 7.” “Ditching 3 and 8.”
It got faster every time. A certain amount of convergence, a certain amount of AI kinda sucks at design speeds up the process. It can make pretty things, it’s not very interested in doing it, but if you shout at it in just the right tone of voice and have a strong idea of what you actually think it can get there. AI really seems to work best with strongly opinionated stubborn people, which is just as well for me.
I’m not a web developer, front end is not my skill set and, whilst I once harboured dreams of a future in graphic design, I’m not an artist. Neither is AI, just to be clear, it’s often referred to as a “force-multiplier” so I was able to create far larger quantities of concepts than I would have ever managed before. I have a pretty short attention span and I was doing this off my own dime so quick iteration and visible progress were key to actually getting something done. Like a lot of folks in tech this wasn’t my first attempt at a site. PhilipWilson.co.uk was purchased about 15 years ago to be a portfolio, then a word-press blog, then a tech demo… This is the first time I’ve got something up that is vaguely complete and not a dumpster fire.
The cassette has the same story, but with less success. Five versions. The first one had screws in the corners and didn’t really look like a cassette so much as a rectangle with reels glued to it. Then I asked for them stacked. Then wider. Then “the pile should spread further out, less and less as it goes.” Then draggable. Then the redesign you’re looking at now: rounded shell, trapezoid relief, the little notched window for the tape. Each version took about ten minutes to land in front of me. If I’d been drawing them myself I’d have stopped at version two and convinced myself it was fine. Getting AI to draw it honestly wasn’t quicker, and it was shitter, some of the cassette iterations were bad, malformed things that my AI chatbot of choice proudly offered up like a well intentioned but deeply incompetent child. Fortunately we haven’t reached the skynet future of our fears so you can skip past the gentle-parenting and get straight into fix-it routines.
A couple of honest things, because the people posting their AI-built sites tend to skip these.
The good ideas were mostly mine, vain I know. The arcade cabinet for the lab section was my call, mid-chat, totally unprompted. The fanned-out pile of cassettes was also mine. Claude had them in a normal vertical stack and I kept pushing until they spread out like a hand of cards. The visual language of the whole site (flat colours, no gradients, slight chunkiness, no neon) came from me vetoing things, not from the model offering them. That’s not to say that Claude didn’t offer things, but they were vague, poorly concepted and kinda muddy. Unguided refinement is not Claude’s strong point. And guided refinement is not most people’s strong point. This is where Slop really hits its stride.
The bad ideas were also mostly mine. I asked for a few things that turned out to be terrible, sat with them on screen for thirty seconds, and went nope, kill it. Which is the other half of why this worked. When you don’t have to draw the bad idea yourself, you stop being precious about it. You just look at it and bin it without the ego shock our sunk cost hitting quite so hard. It wasn’t necessarily faster than me in all places, but even when it spent 15 minutes “finagling…” its way through a problem I didn’t. I sat and used my time in other productive ways, like doom scrolling social media, or making fart jokes to my wife.
The other thing worth saying is that Claude got it wrong constantly. It guessed at what was on my screen instead of checking. It broke the cassette drag more than once. It put up a 404 on the lab route. At one point I had to write “you seem to be guessing rather than actually knowing,” which is a sentence I never expected to type at a computer, and which I now type roughly four times an hour. Giving Claude the ability to verify its own feedback loop is essential to getting it out of the people-pleasing incompetence all AI is hardwired for. Even still it needs strict guidelines and plans to verify against. I find that the more abstract, visual, or opinionated the task: the more likely Claude is to just do weird, useless nonsense. And its desire to always sound confident and affirmative leads to a lot of time wasted.
So if you came here expecting a piece about how AI does it all for you, sorry, haven’t got one. What I’ve got is this: AI lets you be wasteful with concepts. You can throw fourteen homepages in the bin and not feel like you’ve wasted a weekend on each. You can ask “but what if the cassettes fanned out instead of stacking” without committing to twenty minutes of redrawing. You can be picky in a way I, frankly, cannot afford to be picky when I’m the one drawing every version.
This site is exactly as good as I could make it, a version of me that is far more patient and has way more disposable time at any rate. Garbage in garbage out.
The cassette in your hand is draft five. This site has fourteen dead siblings.
That’s what I think AI is actually for.