Minimal UIs
As I’ve vibe-coded more personal projects, I’ve noticed I prefer minimal UIs that leave little room for distraction. Clutter-free interfaces that let you get the work done. I like to push the design to the point where removing any other element breaks the functionality, and I like to leave empty breathing space so thinking can actually happen.
A concept from data visualization that has stayed with me is “saving ink”: do more with less. Less text, fewer colors, less of everything, and more white space. It forces clarity almost without you noticing. This principle guides almost every UI I make.
The first practical decision is what kind of interface actually fits the problem: CLI, website, chatbot, static image, video, interactive visualization, or something else entirely. Start with the one that solves it. Don’t start with the most elaborate option. Relatedly, consider whether you actually have the content to put into the UI. It’s easy to generate layout variations and A/B-test them, but if you lack the time to maintain them or the ideas to fill them, stick to one version. It makes life much simpler.
Clarity of intent is key, and it usually doesn’t arrive until you have something tangible to look at. Over-building UI features at the start is often fine; you need raw material to work with before you can cut it back.
While polishing UIs, it occurred to me that the work is not a waste of time. The better I make what I ship, the more I hope I inspire the people who use it to do their best as well. Attention to detail can compound in ways I don’t fully understand.
Get Evgeny Arbatov’s stories in your inbox
Join Medium for free to get updates from this writer.
Oddly, good UI has not become easier even with AI help. My guess is that it still takes time to realize what is actually right for a given use case. Designing a UI is largely answering one question: what is the most important thing right now? That is how attention works. You can only see one thing, especially on mobile.
I think the ideal state of a UI is that it is never frozen. It keeps evolving based on how you or your users actually access it. A UI that is too hard to change is no good; it is probably the surface most prone to change. The UX should be iterated on often, guided by artistic sensibility rather than A/B tests. The best UIs have an opinion. A UI designed for the average user satisfies no one.
A lot of my better ideas for improving UIs come from conversations with my subconscious while running or walking. A new perspective almost never arrives while sitting at the desk doing business as usual. Something physical has to shift for perspective to emerge.
UIs are best viewed as a conversation. You want to get an idea across and make what you’ve created accessible to other people. You build a UI. It doesn’t work. You abandon it and build another. You either use it, or it sits on a shelf until enough time has passed and you return to iterate.
There is value in going back to projects you already have and making them better — not starting something new, but reflecting on how to make the tools you’ve already built get used more often, even if you’re the only user. Being intentional about updating the UI as your needs change matters.
Is there a use case where a complex UI is justified and a minimal one is not sufficient? Does a minimal approach risk fragmenting the experience into a collection of scripts instead of one coherent whole? Perhaps. But at the scale of one user, you are free to decide whatever works. And when time is short, concise UX beats every feature-rich alternative.
I’m curious what UIs of the future will look like. Given the trend toward personalizing everything, I’d like to see more sites and apps adapt their interfaces to how people actually interact with them. The UI shouldn’t feel static, set in stone. It should evolve in response to user actions: unused features quietly disappearing, frequently used ones becoming more prominent. Everyone ends up with their own interface, unique like their fingerprint.
