I have not written anything in so long that my old Medium account and Blogspot have probably collected enough dust to qualify as archaeological sites.
But since I spent my holiday building a new portfolio website, I thought I might as well bring the blog back to life too. So, how did my five-day holiday go?
Some people would probably guess that I travelled somewhere far away, escaped to the mountains, or spent the whole holiday on a long bike ride.
But noooooo. I stayed at home.
I caught up on some reading, cleaned the house, and rebuilt my portfolio website for what feels like the hundredth time. At this point, redesigning my portfolio might actually be one of my hobbies.
This time, though, I tried a completely different approach. I built it with Codex, mostly by describing what I wanted and arguing with it until everything worked.
I know AI is still a sensitive topic for a lot of creatives. Some people are strongly against using it in design because they feel it takes away from the real creative process, the joy of making things manually, and the human touch behind the work.
And honestly, I understand where they are coming from.
It is a completely different story when someone uses AI to copy another person’s work, generates something in five seconds, and calls it a creative concept. Or when they let AI do all the thinking while they sit there accepting every suggestion like, “Yes, perfect, no notes.” I do not support that either.
But I also genuinely enjoy using AI.
I am not sitting here worried that it is going to take over my job tomorrow. I see it as a tool that helps me move faster and spend less time doing the parts of the process that slowly drain my will to live. It helps me organise things, troubleshoot problems, clean up files, write repetitive code, and handle tasks that would normally take up a large part of my day.
For me, AI is more like a companion or an assistant. Sometimes a very capable assistant. Sometimes an assistant that confidently breaks the entire layout and then acts as if nothing happened.
When used properly, though, it gives me more space to focus on the things that actually matter. The concept, the creative direction, the details, and how the overall experience should feel. It allows me to spend more time thinking about how to become a better creative and produce better work, instead of spending three hours trying to understand why one button has moved two pixels to the left.
I thought rebuilding the website would be much faster this time.
It was not. Noooo.
I have seen people talk about vibe coding and shipping an entire website in four hours, sometimes even less. Meanwhile, mine still took 5 days. There were a lot of iterations. I changed my mind several times, fixed things that were not working, broke a few things that had previously been working, and spent far too long adjusting tiny details that probably nobody else will ever notice.
I also designed and created the visual assets myself. I did not type “make me a portfolio” into an AI tool and accept whatever came out. I still worked on the concept, layout, typography, illustrations, imagery, and overall art direction. AI helped me turn those decisions into a functioning website, but it did not make the creative decisions for me.
The rest was handled by my trusted agents, Dieter Rams, my website developer, and Haugomat, my motion and interaction designer. Designers will probably understand the names. 😉
Yes, I named my agents. Of course I did.
This also means I no longer have to consult my partner about every breakpoint and responsiveness issue, only for the discussion to turn into an argument, me becoming completely overwhelmed, and eventually crying over why the mobile version looks weird. Now I can argue with my agents instead. They never take it personally.
But this is also where a designer’s experience becomes very important. AI can help you build something, but it cannot replace your judgment. You still need to understand design fundamentals, visual hierarchy, typography, UI, UX, and even the basics of motion and interaction.
You need to recognise when something feels wrong, even when you cannot immediately explain why. You need to know when an animation is too much, when the hierarchy is unclear, when a layout is fighting against the content, or when the website has somehow started to look like a tech startup from 2021.
Without that knowledge, the final result can easily feel generic, half-baked, and unmistakably AI-generated. This is why I think designers actually have a huge advantage when using these tools. We already have the eye, the experience, and the ability to guide the output. We can tell the difference between something that is technically working and something that is actually good.
The tool might help with the execution, but the taste, direction, intention, and endless requests to “make it feel slightly better, but not too different” still come from us.
So yes, I understand the resistance to AI.
There are valid concerns around originality, ownership, and the ways these tools are being used. But completely rejecting them also feels like a missed opportunity. Large language models have opened so many doors. They allow creatives to build things that might previously have felt too technical, too expensive, or simply out of reach.
I did not suddenly become a professional developer during my five-day holiday. I also did not manage to ship a website in four hours. But I was able to take an idea, direct the process, make informed design decisions, create my own assets, and build something that I would not have been able to make on my own a few years ago. I even deployed it myself, which made me surprisingly happy. A small win, but I will take it.
AI is not replacing my creative process.
It is helping me spend less time fighting with the boring parts, so I can put more energy into the parts that actually need a human.