Skip to main content
Taste Just Got Expensive.

Brand ·5 min read

AI Made Making Easy. Taste Just Got Expensive.

Everyone can produce now. The advantage has moved to knowing what's worth producing, and what to kill.

Scroll any feed. Half of what you'll pass is obvious AI slop, the six-fingered hands and the empty listicles, and it mostly filters itself out. Nobody loses a customer to garbage; garbage gets scrolled past.

The other half is the problem. The other half is strangely competent. The copy scans. The design is tidy. The video is fine. AI raised the floor of quality for everyone at once, and the ceiling didn't move, so the middle became a crowd. Competent used to be a compliment. Now it's camouflage.

That's the market you're competing in. Not a market short of content, but a market drowning in adequate, where nothing is broken and almost nothing sticks.

The economics did what economics always does

For twenty years the constraint on marketing was production. Making things was slow and expensive, so teams were judged on how much they could ship. Budgets, agencies and job titles were all built around the bottleneck of making.

AI removed the bottleneck. One marketer can now produce in an afternoon what a team used to ship in a quarter. And when something becomes abundant, its value collapses. That's not a hot take, it's the oldest rule in the book. Output is no longer the product. Output is the raw material.

Which raises the question a smart client should ask, so let's ask it: if making is easy now, what exactly are you paying for?

The same thing you were always paying for, whether or not anyone said it out loud: judgement. What to make, for whom, in what order, to what standard, and what to kill even though it took thirty seconds to generate and looks perfectly fine. Production was never the valuable part. It was just the visible part. AI didn't make expertise cheap. It exposed which providers were only ever selling production.

Call the scarce thing taste. Not taste as in beige interiors and expensive fonts. Taste as in the ability to tell the difference between fine and good, and the nerve to act on it.

Taste is a decision, not a decoration

Here's what taste actually is in a working business: it's a filter. The quality of your brand isn't set by the best thing you've ever made. It's set by the worst thing you let out the door. A business with taste isn't the one with the prettiest deck; it's the one where nothing off-brand, off-voice or off-standard ever ships, no matter how fast the machine can make more of it.

Which means taste is mostly the discipline of no. No to the eleventh social post that says nothing. No to the AI draft that's 80% right, because 80% right at infinite volume is how brands dissolve into the beige. The delete key is now a growth tool.

Taste is a system, not a gift

The comforting myth is that taste is something you're born with, one gifted person squinting at things. That myth doesn't scale and it never survives that person's holidays.

In practice, taste gets written down. When we built Spin's own image style, we didn't tell the tools "make it beautiful." We wrote a grammar: one still thing, one moving thing, one colour. Then we tested until we found both failure modes, too gritty on one side, too sterile on the other, and wrote down the middle. Now anyone on our team, and any machine we point at the job, produces imagery with taste, because the taste lives in the system rather than in someone's eye.

That's what a brand platform really is. A voice you can hand to a new writer. Rules a machine can follow and a human can defend. One idea, held long enough to compound. Taste, written down, is the only version of taste that survives scale, and it's the only version AI can actually amplify. Point AI at a business with no written standards and it multiplies mush. Point it at a business with a real system and it multiplies the good stuff.

The machines are judging too

Here's what makes this urgent rather than aesthetic. Your next buyer increasingly asks an AI before they ask a colleague, and AI assistants choose the way careful humans do: they weight distinctiveness, consistency and third-party proof when deciding which brands to name. A brand that looks and sounds like everyone else isn't just forgettable to people. It's statistically invisible to the models trained on people.

So taste now pays twice. It's why a human remembers you, and it's why a machine cites you. Sameness fails both audiences at once.

What to do about it

Run the logo-off test. Put a month of your output next to your competitors' with the logos covered. If you can't reliably pick yours, you don't have a taste problem, you have a taste emergency.

Write your rules down. Voice, image, standards, the failure modes on either side of good. If your taste isn't documented, you don't have taste, you have a person who's busy.

Give someone the no. Every piece that ships should pass one accountable filter. Volume is no longer impressive. Restraint is.

Point AI at production, keep people on judgement. The machine makes more. Only people decide better. Businesses that get this backwards will be extraordinarily productive at becoming indistinguishable.

The machine is now available to everyone, which means the machine is an advantage to no one. What it multiplies is whatever you feed it: your standard.