Back to blog

I Asked for a Better Website. First, I Needed to Decide What ‘Better’ Meant.

3 min read

“Make my website look better” sounds like a reasonable request.

I said some version of it myself. I wanted the site to feel more polished and do a better job of showing who I am.

But that leaves a lot undecided. Better for someone reading a blog? A recruiter scanning my experience? An engineer opening a project?

Before changing the interface, I needed to get clearer about what the site should help people understand.

Seeing options made the decision easier

We explored three versions: a warm editorial layout, a dark product showcase, and a compact profile.

They used the same core facts and projects. But each made a different first impression.

The dark version put the products first. The compact version made experience easy to scan. The editorial version gave the work and writing more room, and felt closer to what I wanted.

I could have spent a long time trying to describe that preference upfront. Seeing the versions beside each other made it much easier to explain.

That's one thing I find useful about building with AI: you can make an idea concrete quickly enough to react to it.

The reaction is often more useful than the original request.

A homepage is only the beginning

Once I chose the editorial direction, I opened the blog.

The homepage had changed, but the writing pages still felt like the old site. The design needed to hold together as someone moved through it.

So we carried the same colours, typography, spacing, and navigation across the articles, project pages, and contact page.

The details changed with the task. An article needed a comfortable reading width and clear headings. A project page needed enough structure to explain the problem, the implementation, and the result.

Repeating the same layout everywhere would have missed that.

Then I removed the photo

The first editorial version included a photo, a short caption, and a line about my stack.

After seeing it on the live site, I decided I didn't want it there.

Nothing was broken. The photo fit the visual direction. I simply preferred the page without it, so we removed the whole block and adjusted the spacing.

That was a small decision, but it captures something I keep noticing about AI-assisted design. You can get a coherent implementation before you've finished deciding what feels right.

You still need room to change your mind.

What I’d do differently next time

I'd start by writing down what someone should understand after a quick visit: what I build, how I think, and where they can explore further.

Then I'd ask for a few distinct directions and compare them against that goal.

AI made trying those directions cheap. Choosing one, noticing where it fell apart, and deciding what to remove still needed attention.

“Better” became useful once I could explain what I wanted the website to do.

aidesignfrontendproduct