Thoughts02 / 02

Yet Another Website

Date
20.11.26
Entry
02 / 02
Reading
5 min
Words
922

I built myself another website. Again. At this point, it's becoming an unhealthy hobby.

Welcome to natts.me, my little corner of the internet. No big plans, no particular purpose. Just a place for my thoughts, things I make, things I like and whatever else comes to mind.

My main website nattawan.com started out on WordPress. Then I rebuilt it with Next.js, React, TypeScript and Tailwind CSS, which, to be fair, was complete overkill. But I wanted to try it, understand it and learn from it. While I was playing around with Next.js and changing things all the time, I started asking myself: do I want to use tools like this for client projects too? As an alternative to WordPress, Shopify, Wix or … wait, is TYPO3 still a thing?

Let's be honest: most people don't want a complicated website. They want one that works, looks good and is easy to update themselves. For that, WordPress is often more than enough. But what if there's another option that fits some projects even better?

Because the problem is always the same: WordPress is well known, popular and easy to use, but it can get slow fast if the site isn't built well. Bricks Builder has improved a lot here, but still. A static website, on the other hand, is super fast, but then nobody can change anything without touching the code.

Astro and Next.js are both fast, but they work in different ways. Astro is made for classic, content-focused websites and sends as little JavaScript to the browser as possible. Next.js, with React and its full-stack features, also works for complex web apps: things with user accounts, dashboards, databases and some kind of world-changing feature. Strava, for example. And me? I just wanted to put some text and pictures on the internet. I probably don't need the kind of setup you'd use to build the next big fitness platform.

I'd been curious about Astro for a while, but I didn't have the right client project to try it on. So I gave myself one: a small blog. This site right here, natts.me.

By the way, I've had this domain for about four years, and it's been sitting around doing nothing. Now it finally has a job.

What I liked most about Astro was the idea of creating static HTML instead of sending a whole JavaScript app just so someone can read three paragraphs. So I started with Astro 7, TypeScript and plain CSS. No Tailwind, no UI framework. Just my own design tokens, a twelve-column grid, some typography and a pretty aggressive acid lime as the accent color. This time, I wanted to be more experimental: glitches, noise, unusual animations, strong colors and small unexpected details, without the whole site looking like a broken arcade machine. Subtle, but not boring.

The first prototype was ready surprisingly fast, and for a short moment, I really believed this would stay a small project.

Then I had the idea to show all the content in fullscreen pop-ups. With their own URLs, of course, plus working browser history, keyboard navigation and responsive layouts. Markdown would have worked perfectly well for the content, but I didn't feel like opening my code editor every time I wanted to write something. So I added Sanity as a headless CMS. Astro pulls in the content during the build and turns it into static pages, while I edit my articles, projects, inspirations and even the little status line in the header in Sanity. And that also solved the problem from the beginning: static and fast, but still easy to update.

I built most of the technical side together with Claude Code. I was in charge of design, decisions and more and more questionable change requests. Claude wrote the code, hunted down bugs and tested everything so carefully, you'd think we were running the systems of an international bank.

And no matter how many guidelines I give it, Claude still finds new ways to surprise me. Not always in a good way. And I've honestly spent days building design foundations and skills. My Notion looks like a technical manual by now.

At one point, the build suddenly showed 59 errors. The fix? One single line of code. Which pretty much sums up software development.

In the end, I have a static Astro website with Sanity, GitHub and Vercel: locally hosted fonts, optimized images, very little JavaScript, and a suspiciously large number of tests behind a pretty minimal interface.

For the record, here's the honest time breakdown:

  • First design + prototype: 3 hours
  • Detailing design + animation: 6 hours
  • Explaining to Claude what "subtle" means: too long
  • Fixing 59 errors with one line of code: 1 minute (plus 2 hours of panic)
  • Finding the right shade of acid lime: don't ask
  • Writing this article: longer than the prototype
  • Writing more articles: we'll see

Could I have just used WordPress or a ready-made Astro template and been done much faster? Of course. But that's boring.

Also, not every website needs to make money. When did we stop putting random nonsense on the internet just for fun? The internet could use a lot more of that, instead of only optimized landing pages, conversion funnels and call-to-action buttons. Nyan Cat understood this more than ten years ago and is still above all criticism.

And after all this effort, I finally have a place to write. Maybe I'll publish three articles here, lose interest and leave the site alone for another four years.

At least that would be consistent.