From Idea to Click: How Claude Code Has Changed My Daily Design Routine
“Cool idea. But that’s way over budget.” I often didn’t even have to hear that sentence. I said it to myself. Long before any idea would have made it into one of my design proposals.
A fancy header animation featuring a spinning globe, or an animation that explains why something just changed on the screen. Good ideas, but time-consuming—which is why they usually ended up on the back burner. “Maybe later.” Later never came.
That’s changed quite a bit over the past few months. I recently built exactly that kind of fancy animation, right in the browser. I’m working more and more with Claude Code and describing in words what I want to try out. I open Figma much less often than I used to. The barrier between an idea and a tangible result has never been lower. This is changing my job as a product designer right now—and for the better.
The design system in Figma was always a step behind
You’re probably familiar with the classic workflow: the design and design system are created in Figma, then handed off to the development team, which translates everything into code. And that’s when things start to drift. A bug fix here, a button there that needs to look a little different on mobile. With every change, the website drifts a little further from the Figma mockup, until—after a few months—no one knows exactly which version is correct anymore.
Replicate the design system in Figma? Theoretically, yes. In practice, though, we lack the time and budget for that. And to be honest, this work also felt pointless. I was recreating components that had long since existed in the code—and were already further along there. The source of truth was never my Figma library; it was always the repository. I just pretended for a long time that it was otherwise.
So now I work where design really comes to life
Today I’m working on the design system directly in the repo. I make sure the components look good and are easy to use. I adjust their behavior, try out different variations, and test them in the browser. It usually starts with a conversation. I describe to Claude what I want to try, look at the result in the browser, and make adjustments as we go: “The spacing is too tight,” or “The animation feels sluggish.” Design emerges more through conversation than through typing.
Does that mean design now bypasses development? No. I initiate changes as pull requests. From there, the development team takes over. The developers decide how to review them and what to accept, modify, or reject. What I build is intended as a proposal that we can use to discuss the actual product.
Click Instead of Thinking
The second difference has less to do with technology than with the feel of the work. In Figma, I have to imagine what a flow feels like. Every click path means extra work. Every connection needs to be defined. And at some point, you end up with more connection arrows than screens. In the browser, I click and scroll my way through and immediately notice where things get stuck. Interaction and motion are suddenly part of the design again, rather than just a note in the handoff saying “should be smooth.”
The globe that would have ended up in the drawer back then
For the Social Income's Transparency Page I built a globe just like that, along with interactive, animated charts. The site clearly shows how much money comes in, how much goes directly to the recipients, and how much is set aside as a reserve.
In the past, I would have drawn this as a static image in Figma and described the motion in Handoff. We wouldn’t have known until development whether the globe rotated smoothly or whether the charts remained easy to understand when interacting with them. This time, I was able to test it in the browser right from the start. There was no “maybe later” this time.
Faster doesn't mean "done"
Now, one might conclude from this that from now on, everything will be built directly in the browser. I wouldn’t. A prototype that runs in the browser quickly looks more finished than it actually is. This has two side effects. I’m more likely to fall in love with the first solution because it already works. And everyone else—including clients—tends to think of it as a nearly finished product.
That's why I try to stick to two simple rules:
- I'd rather have several versions than just one polished one. I use that speed to explore more directions.
- Early prototypes tend to be visually understated. So we can talk about behavior rather than colors and distances.
I'm staying at Figma. Just with a new job.
I still use Figma, but these days I use it for something different than I used to. When it comes to look and feel, branding, color palettes, and typography, I still prefer to explore on a blank canvas. That’s where I can most quickly lay out ideas side by side, discard them, and recombine them.
Clickable prototypes, however, are no longer part of the process. I build them directly in the repo, where they behave just like the real product: with real components and states. For me, it’s therefore less about Figma or code. What I find more interesting is: Where do I explore, and where does the result live? My answer: in both places, but not in the same way. In the end, the result lives in the repo.
We don't know yet where feedback belongs
Not everything has been figured out yet. For example: Where should feedback take place in the future? If customers and users are viewing the prototype in their browser anyway, why are they still leaving comments on Figma screens? We’re currently testing different ways to collect feedback directly within the prototype from the repo. We don’t have a definitive answer yet.
From the handoff to the kickoff
In the past, my work ended with a handoff. Today, it ends with a kick-off. A pull request from me is more like a question for the team: “This is what it might feel like. What do you think?” It doesn’t make me a developer, but our discussions become more concrete because we’re talking about something you can actually click on. And if an idea isn’t any good, we realize it before it gets expensive.
Ultimately, my most important tool is conversation—with Claude, but above all with the team and the people who use the product. If someone says today, “Cool idea. But that’s over budget,” that’s the start of a conversation for me.

Written by
Christian Grutsch
Got a similar challenge on your plate? Let's talk.





