About Project

Every text you will read on my website is written by me. However building this website was a team effort. But not just assembled from a template. The structure, the CSS animations, the layout decisions, the color system - all of it emerged from a back-and-forth conversation with Claude Code and Codex. I described what I wanted, the AI proposed code, I reacted, we iterated. Fast.

When I had my first encounter with LLMs in 2022 to solve a relatively hard logic task on one of my exercise sheets for uni I was amazed as the technology for them had not been older than 5 years back then. I knew that this technology would get better, but had no idea at what pace. However it wasn't until Coding Agents such as Claude Code and Codex came along when I saw the massive opportunities. The energy needed for going from idea to working prototype dropped dramatically. When starting projects, what used to mean hours of boilerplate, trying to finnaly figure out the codebase by reading documentation, and trial-and-error now compresses into minutes.

So for me the speed argument is obvious. But the deeper value is in iteration cost. When trying a different approach is nearly free, you explore more, commit to better solutions, and stop anchoring on your first idea. That's a real shift in how you build.

They also lower the floor for what you can attempt. I can work across domains I'm not yet expert in — write a CSS animation I've never done before, structure an SVG I wouldn't have known to start, wire up a deployment I've only read about. The knowledge gap stops being a wall.

For me personally the risk is building things I don't fully understand. If you can't read what the tool generated and trace through why it works, you don't actually control what you've built. You've assembled a black box. It runs, until it doesn't. And then you have no map.

I don't want to be someone who prompts their way to a result without knowing what happened in between. That's not building. That's delegating all the thinking and keeping none of the understanding. That's not why I study computer science and is not interesting to me. Understanding is the part that compounds.

As a lever, not a crutch. Use AI to go faster on things you already understand, or to get an initial foothold on things you're actively learning. But you should always follow up. Read the output. Ask why. Break it on purpose and see what happens.

The goal is to go from "I could build this" to "I built this" faster, not to outsource the thinking and skip to the end. The thinking is the point. That's where I know that I am growing.

This new technology is probably the most powerful learning accelerator I've encountered. But only if you treat it like a very fast teacher, not a vending machine.

Almost every line of code on this site was generated with help. And I want to understand every line of it, the CSS cascade, the animation timing functions, why the nav blur works the way it does, how the browser paints it. Not because I have to. I probably don't need to know but because I want the mental model, not just the result.

That tension, between moving fast with AI and understanding deeply without it, is something I'm actively working through and that sometimes troubles me. This site is part of that. A place to build, reflect, and document what I actually know versus what I've just seen work.