When production becomes infinite, the bottleneck moves upstream. This is an attempt to peel the onion. To write down what's underneath the word, and why it matters more now than it did five years ago.
When anyone can make anything, the scarce skill is knowing what to make
00 · The Problem
A word used by everyone and defined by no one.
Taste. The word smart people reach for when they want to sound like they understand something they can't quite explain.
It's become the creative class's version of "it just works." A placeholder for the real answer. A way to end the conversation when the conversation gets hard.
I've been guilty of this myself. For years, as a design leader, I used the word taste as a shortcut when I couldn't explain what I actually meant. This is an attempt to peel that onion. To write down how I'm thinking about taste, what's informing that thinking, and to admit, at the outset, that I'm still working it out.
After watching builders work with AI for a few years now, and before that, watching founders bet on products, and before that, storytellers make decisions with a camera, a pattern surfaces. The ones who ship work that moves people don't have a secret sense the others lack. They've learned something specific about judgment, and they can name the parts. The ones who freeze have confused taste with mystery, then borrowed the confusion to avoid doing the work.
This is an attempt to decode it. Not into a feeling, but into a framework. Two of them, actually.
01 · The Taste Economy
When production becomes infinite, the bottleneck moves upstream.
Something changed between 2023 and 2026 in the way builders work. I watched it happen across projects I was part of and projects I was simply watching. The cost of production collapsed. Code, images, copy, UI, music, video, all of it cheaper and faster than it had ever been. By most estimates, over 90% of new online content in 2026 is generated by a model.
That isn't a trend. It's a phase change.
When everything can be made, the pressing question is no longer can we make this? It's should we make this? The second question is a different species. It doesn't yield to effort. It doesn't yield to better tooling. It yields only to judgment.
That second question is a question of taste.
Anu Atluru said it cleanly: "In a world of scarcity, we treasure tools. In a world of abundance, we treasure taste."[01] She's right, and I think the implication is larger than most people have followed through on. The bottleneck in any creative or strategic pipeline has quietly moved from execution to judgment. Not for some teams. For all of them.
This is what I've come to call judgment under abundance. When everyone can make anything, the scarce skill is knowing what to make, and what to throw away before it reaches another human.
Judgment, at its highest expression, is taste.
Features can be copied in a week. Functionality can be matched in a month. What can't be copied is the feeling of something made with intention, built by someone who knew what they were doing and why. That feeling is what the word taste is pointing at. And it's what AI, for all its brilliance, can't fabricate on its own. Rick Rubin put it most cleanly: the AI doesn't have a point of view.[06] It will generate anything you ask for. It can't tell you which of the things it generated is worth asking for.
That's the job. It's always been the job. It's only now, with the production layer collapsing, that the job has become visible.
02 · What This Essay Is
One phenomenon, two frameworks.
Most writing on taste answers only half the question. It tells you taste matters, and maybe that taste is learnable, and then it leaves you there.
I want to do two things instead.
First, I want to decompose taste. To answer the question what is it made of? The answer I keep returning to, tested across a year of watching AI products ship, is a four-part one: conviction, experimentation, audience, and context. This is The Compound, and it's the first framework.
Second, I want to scale taste. To answer a question most essays refuse to touch: how does personal taste become a team's advantage? Because if taste only ever lives inside one head, it dies when that head leaves the building. This is The Reach, and it's the second framework. It emerged from watching organizations try to hold a coherent point of view while the ground shifted beneath them, through acquisitions, reorgs, leadership transitions, and the arrival of AI tools that multiplied every team's output overnight.
You can read either framework first. They answer different questions. Together, they begin to make taste tractable. Something you can build, measure, teach, and defend.
This is written from an AI builder's vantage. From a practitioner who came through design, built products, and now ships AI at scale. The frameworks aren't design-specific.
When I say AI builder, I mean something specific. Not the person who wrote the model. The person who ships with it. In 2026, that person is often a collapsed role. Part engineer, part designer, part product manager. Someone who uses AI to go from intent to artifact faster than any of those disciplines could alone. An evolved PM. An evolved designer. An evolved engineer. The edges of the old roles blur where the model sits.
Which is why this isn't written only for designers. If you build product, the ingredients translate to roadmap calls and feature conviction. If you write, they translate to sentences kept and cut. If you run marketing, they translate to campaigns shipped before the focus group agrees. The discipline changes the artifact. It doesn't change the anatomy of the judgment. That's the premise this runs on.
03 · The Builder's Burden
A leader's taste is visible whether they want it to be or not.
What a leader ships is a reflection of how they think. Whether they want it to be or not.
If a leader's taste lives only inside their own head, here's what happens. The product gets fuzzier in every handoff. The principles on the wall fail to survive the first tradeoff. The strong designer they hired last quarter ships something they never would have approved. Not because that person has bad taste, but because the leader never made their taste transmissible.
This is the builder's burden in the age of AI. The volume and velocity of decisions has gone up. Every team now ships more, faster, with less review. The PM shipping three specs where they used to ship one. The writer producing a hundred headlines where they used to draft ten. The marketer running forty concurrent variants. A leader who can't cultivate taste in others, not just exercise it themselves, becomes a bottleneck. Or worse, a ghost haunting the product with decisions they didn't actually make.
The real test of a leader isn't whether they can exercise taste. It's whether the people around them start making decisions the leader would have made, without the leader in the room.
Taking taste out of the leader's head and turning it into something you can see, govern, and scale is the move the next framework attempts. Ingredients first. Reach second.
04 · What Follows
Two frameworks. One argument. Taste is structure, not accident.
A tour through both frameworks. First the compound (what taste is made of), then the reach (how taste scales). A chapter on practice. A chapter on sources. Read in any order. Read the one you need.