I could never get a website builder to do what I wanted

Access to tools was always the actual bar. Not talent.

A freelancer posted on LinkedIn recently asking a fairly ordinary question: what website builder do you use for your own site? Cost and ease were the priorities. Nothing fancy. Just something that looked good and didn’t take three weeks to figure out.

Sixty-odd replies came in. Squarespace, Wix, Webflow, Framer mostly.

My answer wasn’t a website builder at all. It was Claude Code and ChatGPT, some vibe coding, and a GitHub repo.

I could never make WordPress do what I wanted

I’ve never been a designer, and WordPress has never let me forget it.

The themes always needed a plugin I didn’t have. The plugin always needed a setting I couldn’t find. The page builder promised drag and drop, and then punished you the moment you dropped something one pixel out of place.

I tried the simpler web builders too. Same problem, smaller scale. None of them were short on features. What they couldn’t do was exactly what I asked for. There was always a compromise buried in the template, a layout that almost worked, a setting that got you most of the way there and stopped.

So the honest options were: spend a weekend fighting software and end up with something that looked like I had, or pay a company to do it properly.

Most freelancers and job seekers I know picked neither. They picked a bad LinkedIn profile and called it done.

What actually changed

I wasn’t describing a one-off experiment in the replies. I’d already done a harder version of the same thing on the site you’re reading this on — not a five-page brochure site, a full editorial system, built the same way. If the tools can hold that together, a freelancer’s portfolio site is the easy case.

I added a bit more further down the thread:

“If you get the hang of the tools, you can create pretty much what you want. The toughest bit can be the hosting, but if you’ve used AI tools before, you’ll know it can pretty much help you step by step.”

The tools that actually got me there were Claude Code and Codex, working alongside each other. Not because the web builders were short on options. Because they still couldn’t do exactly what I asked for, and this time I finally got something I was genuinely happy with, not something that would do.

That’s not a story about a clever new tool. Website builders have existed for two decades. The tools were never really the limiting factor.

What was limiting was the amount of translation required between an idea and a finished thing, and who was expected to supply that translation themselves.

Not everyone was trying to become a designer

I’ve made a version of this argument before, about writing. It applies here too.

Most people who need a website were never trying to become web designers. They’re freelancers, consultants, people between jobs, people running something small on the side. Nobody asked them to develop taste in typography or spacing. They were asked to show their work, and a website became the way you’re expected to do that.

Telling them to either learn design or pay someone who already has it was never really a meritocratic bar. It mostly sorted people by budget and free time, and called the result quality.

Now a much larger group of people can close that gap themselves. Not because they suddenly acquired design skill. Because the tool absorbed the part of the job that used to require it.

This doesn’t make everyone a designer

I want to be careful here, because the easy version of this argument overreaches.

Someone else in that same thread pushed back a little: a tool like Framer was genuinely good, but they’d have struggled without a full-stack developer in-house. That’s a fair caveat, not a contradiction. The floor has dropped. It hasn’t disappeared.

Genuinely good design is still a skill, and it still matters more in some places than others. A brand that depends on how it looks and feels probably still needs someone who has spent years developing an eye for that. Cheap generation is also going to produce a lot of websites that look like every other website generated the same way, for the same reason cheap writing produces a lot of writing that reads like every other cheap writing.

That’s a real cost. I’m not pretending it isn’t.

But most of the freelance websites that never got built weren’t failing to become award-winning design work. They were failing to exist at all, because the person who needed one didn’t have the skill, the time, or the budget to clear what used to be a much higher bar just to get something functional online.

The pattern keeps repeating

Writing, design, whatever comes next. The shape of the argument doesn’t change much.

The barrier was rarely the ceiling. It was the floor — the amount of skill required just to produce something adequate, before anyone got anywhere near good.

AI is mostly collapsing the floor, not replacing the ceiling. People with real design judgement, like people with real writing craft, still have something the tools don’t supply on their own.

But a lot of freelancers and job seekers were never competing for the ceiling. They just needed the floor to be low enough to stand on.