I’m not gonna lie. I judged it before I used it.
Not because I tested it and it failed. I didn’t even get that far.
It was the UI.
The colors, the fonts, the desktop app… it all felt muted. A little old. Not nostalgic. Just dated in a way that didn’t make me want to spend time in it.
And yeah, I know that’s shallow. But I’m a designer. That’s literally how the evaluation starts whether I want it to or not.
So I side-eyed it.
Especially because I was already using ChatGPT. Solely based off of appearances, it was the much better app. I liked the way the fonts looked, colors weren’t abrasive. Hell, I was even paying for it. It was already in my day-to-day workflow.
Claude wasn’t part of that system.
But I tried it anyway.
Not as a switch. Not as a statement. Just curiosity. Low stakes. Small prompts. Nothing important.
And it didn’t fall apart.
That was the first thing I noticed.
No friction. No weird misunderstandings. No constant re-explaining what I meant. It just held context and stayed with the direction I was already thinking in.
And the way I think is not linear—it’s structure, edge cases, flow, “what happens if this breaks, what if this scales differently,” all happening at the same time.
Most tools require you to flatten that into something simpler before they can respond properly.
This didn’t force that as much.
It just kept up.
So I kept using it.
Then I noticed I was reaching for it first.
Then I noticed I wasn’t really going back to ChatGPT for certain types of work.
And now it’s basically my default until I hit usage limits for the week.
Which is interesting, because I fully expected ChatGPT to be the long-term tool in my stack. That was already the established behavior—paid plan, integrated workflow, no reason to switch.
And I also fully expected I wouldn’t like Claude, because again… the colors. Designer problems.
But function doesn’t care about first impressions.
And eventually, neither did I.
Now it’s just part of the system I work in.