Lately I have been heads down building a tool/service, called Traction, for founders/entrepreneurs, and newsletter operators (because they’re entrepreneurs too) that helps them employ their voice across the socials-sphere in order to help their products/services/newsletters gain traction. As someone who once coded software for a living, the new AI coding tools, like Claude Code, make the entire process magical. Like…truly and completely magical!

And yet…I’m left wondering: even though I can make the thing, should I make the thing? In other words, what’s the problem/pain I am trying to solve? Is it big enough and pervasive enough that I should solve it for me + others? Even if it is a real/big problem, how best should I solve it?

This is where design and taste come to play! And…as I wrote in my piece on taste a few months ago, I think that ability to recognize problems, opportunities, quality and make better decisions is something we develop over time. The more I think about AI Compression, though, the more I think taste is moving upstream. Less toward the artifact itself and more toward deciding which problems deserve to be solved in the first place.

Read on below…

— Justin Lokitz

Per the above, I've been thinking a lot lately about where taste actually shows up in the work, especially as AI makes more of that work easier to do. A few months ago, I wrote about taste as something we develop over time through exposure, reflection, comparison, and making. My argument was (and still is) that taste isn't some magical gift reserved for a handful of especially creative people. It's a practiced ability to recognize quality and, perhaps more importantly, to understand why something is good.

But And…I've started to think AI is changing where that ability matters most.

For most of the history of digital product development, execution was the constraint. Building software was expensive. Prototypes took time. Design changes could trigger weeks of engineering work. As a result, a metric shit-ton of organizational energy went into figuring out how to get things made. I can viscerally recall my early years as a product manager at Autodesk, spending weeks writing product requirement documents (PRDs) and full years working with huge development teams, whittling down my vision set forth in those PRDs to “what’s possible” this year vs. next year.

Now, a single person can generate interfaces, build prototypes, test multiple directions, and sometimes ship an entire product in a fraction of the time it would have taken a small team a few years ago. Crazy!

This is essentially the AI Compression many of us, me included, have been writing about. Design, product, and engineering are beginning to collapse into a much tighter loop, and the distance between an idea and something real is getting smaller. But…when execution gets easier, it doesn't necessarily make the work easier. It moves the difficult decisions somewhere else.

I was thinking about this while listening to Tony Fadell (on Lenny’s Podcast) talk about how he decides what is worth building. Fadell, who helped create the iPod and iPhone and later founded Nest, said he always starts with pain. He looks for a persistent human problem and then asks whether some new technology has emerged that allows us to solve it in a meaningfully different way.

The order is important because much of what I see happening with AI today works in reverse. We start with a new model or technical capability, build something because we can, and then search for a customer who might want it. The result is a growing collection of technically impressive products solving fairly unimportant problems. BTW: I am complicit in this as well. AI tools feel magical. Yet that magic also has the effect of dazzling (see the Twilight book series for this definition) the builder.

Fadell also talked about the debate inside Apple over whether the iPhone should have a physical keyboard. The team spent months testing typing speed, error rates, hardware, software, and every variable they could reasonably measure. Eventually, though, the data stopped being useful. It couldn't definitively tell them which direction was right. Someone had to make a judgment call.

That’s the key! And…this is where I think taste is moving.

We've traditionally associated taste with the things we can see and touch: the interface, the interaction, the typography, the details that make one product feel considered and another feel generic. Those things still matter. But AI can now produce an almost infinite number of plausible versions of them, which means our ability to generate options is rapidly becoming less valuable than our ability to choose between them.

Increasingly, taste is being exercised much earlier in the process. It's deciding which problem is actually painful enough to solve. It's recognizing that a new technology changes what is possible before the market fully understands it. It's knowing which three features matter and having the confidence to remove the other twenty. It's seeing the product as part of a larger system that includes the story, pricing, onboarding, distribution, and the context in which someone encounters it.

Those are all acts of judgment, and there is rarely a spreadsheet that can make those decisions for you.

This is also why I'm increasingly skeptical of the idea that AI will reduce the importance of designers. It will absolutely reduce the value of some of the things designers have historically been paid to produce (pixel pushers beware). I think we'd be kidding ourselves to argue otherwise. But the people who are good at understanding humans, synthesizing complexity, framing opportunities, making tradeoffs, and recognizing quality before it becomes obvious may will become considerably more valuable.

Fadell said something in the interview that I keep coming back to: technology is in service of the customer, not the other way around. That seems painfully obvious, and yet the AI economy is currently producing a lot of technology that asks humans to figure out why they need it.

As execution continues to compress, I suspect the most important design skill won't be our ability to make more things. It will be our ability to decide what deserves to be made in the first place.

And I think that may be where taste matters most.

Subscribe to Design Shift for more conversations that help creative professionals grow into strategic leaders.

No personalized Post-It Note this time as I’m traveling. Sorry!