For a while, I felt a little strange about using AI to build websites.

There was always this question in the back of my mind: If AI helped me build it, how much of it is actually mine?

I think about that differently now.

AI has become part of the way I work. I use tools like ChatGPT and Claude Code to explore ideas, experiment with different directions, solve problems, and get things moving when I don't know exactly where to start. It has changed how quickly I can go from an idea to something I can actually see.

But it hasn't changed the part of the process I care about most.

Deciding what feels right.

AI can build something. I still have to make it good.

One of the interesting things about AI-assisted website building is how quickly you can get a result.

You can describe an idea, generate something, look at it, and suddenly there's a website sitting in front of you.

And sometimes it's actually pretty good.

Then you look at it again.

Something feels off.

Maybe the spacing doesn't feel right. Maybe the typography doesn't have enough personality. Maybe there's too much happening on the screen. Sometimes everything technically works, but the website still feels like something you've seen before.

That's usually when I start changing things.

I move things around. I remove sections. I change the layout. I try a different direction. Sometimes I keep almost nothing from the first version.

And I think that's an important part of using AI that doesn't get talked about enough.

Getting something generated isn't the same as knowing what should stay.

More ideas don't always make things easier

Before using AI, I might have had a few ideas for a website.

Now I can have dozens.

At first, that sounds like an advantage. And it is, in some ways.

But having more options also means having more decisions to make.

AI can show me ten different directions in a short amount of time. But it doesn't automatically know which one fits the project best. It doesn't know which detail makes the design feel like mine, or when an interaction has gone from interesting to unnecessary.

I still have to make those calls.

That's probably where my creativity matters the most.

I'm not trying to compete with AI at generating things. It can do that much faster than I can.

I'm more interested in developing the ability to look at something and think, “This isn't quite right yet.”

Then figuring out why.

I don't want to hide that I use AI

I used to think I should probably keep AI out of the conversation.

There was this feeling that if I told people I used AI to build websites, they might think I didn't actually do the work.

But I don't really believe that anymore.

Tools have always changed the way people create things.

The interesting part isn't whether a tool was involved. It's what you do with it.

If something helps me explore an idea faster, I'm going to use it. If it helps me get past a problem that would otherwise slow me down for hours, I'll use it. If it gives me a starting point that I can improve, that's useful too.

I'm not interested in pretending my workflow is something it isn't.

I use AI.

I also make a lot of decisions myself.

Both things can be true.

The part AI can't decide for me

The more I use AI, the more I realize that the final result still comes down to taste.

Should this section even exist?

Does this animation add anything?

Is this image helping the page or just filling space?

Does the typography feel intentional?

Would the website be better if I removed half of it?

Those questions don't disappear just because AI can generate a website.

If anything, they become more important.

Because when building becomes faster, taste becomes more visible.

It's easier to make something.

It's harder to make something that feels genuinely considered.

AI changed my workflow, not my reason for building

I don't use AI because I want to avoid learning or thinking.

I use it because I want to spend more time actually experimenting.

I can try an idea without worrying that I'll spend an entire day just getting the first version working. I can explore different possibilities and then focus my attention on the parts that matter to me — the visual direction, the experience, the details, and the overall feeling.

That's what I like about my workflow now.

Framer gives me a visual environment where I can shape websites, while tools like Claude Code and ChatGPT help me explore and build in different ways.

Sometimes I use one approach.

Sometimes another.

Sometimes I start with an idea and end up somewhere completely different.

And that's okay.

I'm still figuring it out

I don't think I've found the perfect way to build websites.

I'm still learning.

I'm still experimenting with Framer. I'm still figuring out how I want to use vibe coding. I'm still discovering what kind of websites I enjoy making and what kind of creative direction feels most natural to me.

But I don't think I need to have all of that figured out yet.

Right now, I'm enjoying the fact that I can have an idea, build something from it, look at it, question it, change it, and try again.

AI has made that process faster for me.

It hasn't made the creative decisions for me.

And honestly, I think that's the part I want to keep.

AI can help me build faster. It can't decide what I want to build.