Taste is all you need

Building software these days is way too easy.
I see a market flooded with "solutions", digital products and landing pages that pop up overnight. And almost all of them smell the same. They smell like Claude.
Why is this happening? Because without taste you don't notice you're shipping exactly the same thing as everyone else. You just accept whatever comes out.
Building is too cheap
Today I design, prototype in code, test, document, write copy and do QA. I validate in a day what used to take me weeks. A Claude subscription, even the most expensive one, is ridiculously cheap for everything it lets you do.
The problem isn't the tool. It's that whatever comes out by default looks like everything else that came out by default.
With this democratization of code, value shifts to deciding with taste.
What does having taste mean?
Taste isn't just aesthetics. It's not about having a good eye for combining typefaces, although that helps.
To me, taste is:
- Knowing what not to build. The most valuable decision in a project is often the feature that never gets built. A lot of my day-to-day work is saying no. It took me years to recognize this, and it has even cost me some grief and my place on a project. Not everyone is ready to hear that their wonderful idea makes no sense.
- Noticing that what just came out is the same as everyone else's. And not accepting it as is.
- Defending a decision in front of a stakeholder. With business, user and brand arguments. These days I say "I like it" less and "it makes sense" more.
- Knowing when an ugly MVP is enough. You don't need a custom design to validate an idea. And yes, that's a designer saying it.
None of that comes from a prompt. It comes from shipping things, getting them wrong and seeing how people actually use your software.
AI gives you speed, not direction
Every week I read that designers are dead. And every week I laugh a little.
I use AI every day, and that's exactly why I know what it does: it amplifies what I already know. If you know where you're going, it gets you there faster. If you don't, it gets you nowhere faster.
The data points the same way. According to McKinsey's The state of AI in 2026, 80% of respondents say AI has improved their productivity. But only half say it helps them make better decisions.
Producing more is easy. Deciding better, not so much.
AI has democratized production. Taste is still ours.
How I train my taste
Taste can't be improvised, but it can be trained. This is what I try to do day to day:
- I question the first output. Whatever comes out by default is a starting point. I add something personal, think it over and do it again.
- I validate before scaling. I prototype fast, put it in front of users and let the data argue with my gut.
- I force myself to explain the why. If I can't justify a decision in two sentences, it's probably not a decision. It's a default.
- I cross disciplines. Taking my designs from Figma to code has made me a better designer. I understand the limits, the costs and which details really matter.
What really matters now
AI gives you speed. But without taste you won't get very far. That's exactly what separates just another product from one people remember.
If you're building digital products, surround yourself with people who have taste. And if you're a designer, relax: there's plenty of work. But the work isn't producing anymore. It's deciding.



