There’s a term now for a very specific kind of ugly: “AI slop.” Purple gradients. The font Inter. Three icons in a row, each with a one-line caption underneath. It’s instantly recognizable, and it’s not really an aesthetic — it’s a symptom. Large language models trained on millions of scraped Tailwind CSS tutorials output the statistical median of that training data, and the median turns out to be purple and forgettable. But here’s the part that matters more than the AI angle: this exact failure mode existed years before any of these tools did, and it has nothing to do with artificial intelligence specifically. It’s what happens any time a design decision gets made by asking “what’s popular right now” instead of “what actually belongs to us.”
The Mechanism Doesn’t Care Who’s Doing the Copying
Optimize for broad appeal and you get pulled toward the average, whether the thing doing the optimizing is a nervous marketing team studying competitor screenshots or a neural network reproducing patterns from its training set. The mechanism is identical. A human designer flipping through Dribbble looking for “what’s working right now” and an LLM sampling from its most common training examples are running the same query against the same kind of corpus: what does everyone already like. The answer to that question is, by definition, never going to be distinctive, because distinctiveness is the opposite of consensus.
This matters for brands specifically because a brand’s entire job is to be recognizable as itself. A visual identity built from “what’s trending” is structurally incapable of doing that job, no matter how well-executed the trend is in the moment. It’s borrowing its personality, and borrowed personalities expire the second everyone else borrows the same one.
The Illustration Style That Ate Big Tech
Facebook found this out in a particularly public way. In 2017, the company rolled out “Alegria” — Spanish for joy — an illustration system built by the studio BUCK: flat, colorful human figures with long bendy limbs, tiny heads, and skin in unnatural blues and yellows, always mid-motion, always looking cheerful about something. It worked. It made a company that most people found faintly sinister look approachable and fun.
Then everyone copied it. Slack picked it up. So did Google, WeTransfer, a pile of banking apps, and what felt like every startup that needed to fill white space on a homepage without hiring a photographer. It was cheap, it was easy to animate, and — this is the detail that matters most — one of BUCK’s own designers later said the style was originally more intuitive and complex, but got simplified during internal review specifically to make it easier to replicate. The simplification that made it spreadable is the same simplification that made it disposable.
By 2021 the backlash had a name — “Corporate Memphis” — and then several meaner ones, memes, a dedicated subreddit, a parody of Goya’s “Saturn Devouring His Son” made from the style’s own long-limbed figures. A journalist covering the collapse put the core problem better than I could: the style made Big Tech companies look friendly and human-scale, which was more or less the opposite of what those companies actually were. By 2023 even Netflix was pointedly moving away from it. The style had become a tell — evidence that a company had reached for the same visual shortcut as everyone else instead of figuring out what it actually looked like. As one critic summarized it, the whole style “represents everyone and speaks to no one.”
Nobody at Slack or WeTransfer or any of the banks set out to look identical to their competitors. They each independently made the same reasonable-sounding call — this is working for a big, credible company, so it should work for us too — and the aggregate of a hundred reasonable individual decisions was a visual language with no author left in it.
There’s also a market angle that gets skipped in most retellings of this story. Illustration libraries like Humaaans and subscription services selling pre-drawn Corporate Memphis figures for a monthly fee sprang up specifically to serve companies that wanted the style without commissioning original work. That’s the tell that the whole thing had stopped being a design decision and become a commodity purchase — a checkbox next to “modern illustration style,” priced like stock photography, because by that point it functionally was stock photography wearing a trend’s clothes.
The Company That Said “We Set the Trends, Man”
For contrast, look at 37signals, the company behind Basecamp, which has been publicly, almost stubbornly indifferent to design trends since 1999. Writing about one of their own site redesigns years ago, the team put it about as bluntly as a company ever has: “We tend not to follow the trends. WE SET THE TRENDS, MAN!” It reads like a joke, and it’s partly meant as one, but the underlying practice is real. Their internal standard for every design decision has always been some version of “clarity over cleverness” — does this help the person reading it understand what we do, full stop. Not: does this look like what other credible software companies currently look like.
That discipline is why a company that’s been making roughly the same style of plain, text-forward, occasionally odd-looking product for over two decades still reads as having a point of view, while entire cohorts of Corporate-Memphis-era startups now look like a specific, slightly embarrassing snapshot of 2019. Basecamp’s design was never trying to signal “modern.” It was built to be legible to actual users, which happens to age fine, because usefulness doesn’t really go out of style the way visual fashion does.
What This Actually Means for a Design Decision
None of this is an argument for weirdness for its own sake — plenty of deliberately strange design is just a different flavor of trend-chasing, aimed at a smaller, more self-congratulatory audience. The actual test is simpler: can you explain why a design choice serves this specific brand, in this specific moment, for this specific audience — or is the honest answer “because it’s what’s working right now”? A digital branding process that starts from the second answer is, structurally, doing the same thing an LLM does when it defaults to purple. It’s sampling the median and calling it a decision.
This is a genuinely hard thing for a design agency for startups to push back on, because “make it look like the thing that’s currently getting funded” is a comfortable, defensible request, and “figure out what’s actually true about this company first” takes longer and produces something that might not test well against this month’s Dribbble front page. But the trend always moves on, and whatever replaces Corporate Memphis will get replaced by something else, on schedule, forever. A design built from an actual conviction about who the company is doesn’t have that expiration date, because it was never renting someone else’s popularity in the first place.
The real question worth sitting with isn’t whether your brand’s current look is trendy. It’s whether you could explain, in one sentence, why it’s yours and not just borrowed from whatever’s winning this year.
