When Building Becomes Default

Let me tell you what happened to me this last month.

For years, I've been a heavy user of no-code platforms. This website was built entirely in Webflow. I also played with Framer. Before that, I had several WordPress installations, some self-hosted, some on third-party services. And long before all of that, I was building sites completely on my own, writing the code myself. I've always been an early adopter of frameworks and services. Curiosity has always pushed me toward whatever was new. I've tried most of the spectrum.

Two years ago, I redesigned this site in Webflow with a completely custom design. I spent almost a month tweaking animations, layouts, and transitions. It was genuinely fun. It reminded me of the early days of Dreamweaver (jezz I'm old), when building for the web still felt like crafting something by hand. Webflow made things that would have been slow and painful in pure code feel smooth and almost playful.

But the trade-offs were always there.

• The monthly cost.
• The add-ons.
• The subtle feeling that the site wasn't really mine.

If I wanted a new locale, that was an upgrade. If I needed a specific feature, that was another decision tied to a plan. I knew all of this when I signed up. I wasn't naive. I was paying for convenience and for peace of mind.

And to be fair, peace of mind matters. Years ago, I hosted multiple WordPress sites on my own server. One of them got hacked, and I lost everything. It's the kind of experience that makes value-managed services.

So I stayed, and it worked for a while. But then AI code assistants became actually usable.

This summer, I built a small app for myself using Codex and Claude Code. Nothing revolutionary, just something I needed called Digital Drawer. But the process felt different. I wasn't fighting a platform. I wasn't adapting my idea to fit someone else's product logic. I was just describing what I wanted and iterating.

That's when the shift happened this time regarding my websites. It wasn't about saving money. It wasn't even about speed. It was about not depending on someone else's roadmap.

So I decided to port this entire website away from Webflow. It took me about a week of pretty intense work, especially since porting is always messier than starting from scratch. Also, I took the opportunity to create a Markdown system instead of using a database for the articles, since this blog doesn't need one.

But once it was done, something felt different. I understood every piece of it. I could change anything. I wasn't constrained by pricing tiers or feature gates. Also, the code, in a GitHub repository, I remembered the wonderful feeling of having control of your code. Somewhere along the way, we gave up a bit of that freedom. We traded it for convenience. We added dependencies on services, platforms, distribution algorithms.

That's exactly why I created my own blog in the first place. I like owning my own space instead of publishing only on Medium, LinkedIn, or X. I’ve used all of them. They’re useful. But they’re not mine. With the code back in my hands, and with AI assistants making complexity manageable again, that freedom came back.

Also, the other day, something small happened that was also powerful in its own way. I posted this article about Predictable Machines. And, on Bluesky, a friend asked if I had an RSS feed.

Although there is probably some service that has it by default, because RSS is kinda standard, that question would have meant checking if my plan supported it, looking for a workaround, or adding some third-party integration. This time, I opened Claude Code and asked it to implement RSS. Five minutes later, it was live.

It was just RSS. But it represented something bigger. I didn't need permission. I didn't need to upgrade. I didn't need to wait for a feature request to be approved.

I have the feeling this is only the beginning. No-code platforms won't disappear overnight, but the reason they exist is shifting. When building custom software becomes accessible enough, many SaaS products that are essentially wrappers around workflows start to look fragile.

If you can describe what you want clearly enough, you can build it. And once you realize that, it's hard to ignore.

I ported four websites so far (one of them, Open Source btw: The Margaret Hamilton Project), and the number is growing.