As always with new things or even trends: battle lines are drawn! Either it’s lavishly praised to the heavens by the same people in my industry as the ultimate solution to everything. And then there are those with total rejection, because it’s absolutely terrible for this and that reason.
The truth, as so often, lies somewhere in the middle, and I generally tune out the extreme positions just mentioned, because they surprisingly often come from people who’ve never seriously engaged with it. Regardless of which position they hold.
The problem with total rejection is often the doom-mongering of the critics. But there’s rarely a deep black or a pure white. It’s usually one of a thousand shades of grey.
1000 shades of grey, or what?
Years ago, I picked up front-end fundamentals, can write a bit of HTML and CSS, and understand a little about how JavaScript works. When I see website code, I don’t necessarily see whether it’s good, but what it’s trying to achieve.
I reckon that’s a difference when I tackle something and vibe code, compared to someone who hasn’t got a clue about front-end development. And there we are in a grey spectrum!
Enough justification!
My old website at wirsindfein.de served me for a very long time as a business card for my freelance work, which I no longer pursue since the end of 2022. The site had perhaps got a bit tired, or rather like an old, albeit quality, jumper. Still looks good, still fits, no holes… but worn a thousand times!
It was time for something new!
My first vibe-coding test run was this small site: eichhof.co
A placeholder page that I use for domains that I only have for email addresses, for example. Doesn’t need to do anything, but perhaps someone might stumble upon it.
At some point, I had the idea for the particle smiley and gave it a go with Claude, and it was surprisingly quick to make initial ideas immediately visible and functional.
Got a taste for it!
After that, the door was wide open. I wanted to tackle wirsindfein.de, but also shed more than just the old jumper. I don’t need a site that showcases my skills for potential clients, I just need something I can use on my profiles across the web that briefly shows who I am.
An about-me page with a new URL
I started with the circular photo, the text, and the buttons. Then it became – completely without reason – multilingual. Then came more buttons, which also change their order depending on the language version. The email addresses needed to be as unrecognisable to bots as possible. The first Easter egg was built in, an imprint in a pop-up had to be added too. Why not a dark mode as well? Oh, another Easter egg, please. Shall we cram the thing with structured data as well? And the text links need a preview with self-updating screenshots of the pages.
There’s a disproportionate amount of functionality in this small site, and that’s precisely what made it fun. Quickly visualising an idea and potentially developing it further, adjusting it, scrapping it. Just doing it!
Last week, I then turned the site into a GitHub repo. Also from within Claude, with direct integration, so that what I tell the machine gets pushed straight away and through another connection goes live immediately. In between, I had Claude run code and SEO audits itself. Because redundancies often creep into the code when making ad-hoc changes.
That’s where we get to the difference it can make when you consciously engage with such a tool. Vibe coding initially means: I chuck my ideas into an AI and take what comes out. That needn’t be, and probably won’t be, good yet. But it doesn’t have to stop there.
I had the code checked by other AIs as well. Simply to get a feel for how they react. Often superficial rubbish came out that you first have to recognise as such. Ultimately, though, the “audits” in Claude itself worked best.
If you critically examine the output to the best of your understanding, and you can weigh up a good overall picture, it’s quite possible to produce optimal quality in extremely short time with vibe-coding.
I’m really pleased with my result, and you can certainly argue about why inline CSS and JS (because of the number of requests) and (now completely refactored) which function is implemented how well. Does it even need JS and so on? I reckon that’ll largely just be down to individual taste.
Leave a Reply