30/05/25 – BY APRIL BUTTEN
What AI can’t brand
Why originality is being replaced with optimisation, and what to do about it
Let’s be honest. Most of us didn’t start using ChatGPT because we were tech-optimists. We used it because we were drowning. Building a brand from scratch means spinning ten plates with two hands, and if a chatbot could edit your pitch deck while you answered a supplier call? Done.
That was the appeal. Cheap, fast, competent enough. So we used it. And yes, we still do. Daily. For research. For structure. For image testing. Whisper transcribes our 90-minute calls so we don’t have to scroll through pages of notes.
But here’s the moment we noticed the shift: when brands started skipping the hard parts – tone of voice, naming, founder story – because a prompt could give them a B+ version in under 10 seconds. Not great, not terrible. Just… fine.
And fine is where great brands go to die.
What’s actually at risk
There’s this narrative floating around that AI is killing creativity. But that’s not quite it. What it’s actually killing is originality.
Because most of the time, AI isn’t being used to push ideas further. It’s being used to avoid having an idea at all. Brands are launching with names the founder doesn’t love. With tone of voice that reads like a trending post. With campaigns that feel high-production, but hollow.
It’s not that the work is bad. It’s that it has no spine.
The AI-generated brand isn’t a disaster. It’s a decoy. It makes you feel like you’ve done the work of branding when all you’ve really done is style the surface.
The real divide isn’t human vs. machine
It’s founder-led vs. filler-led.
You can spot the difference immediately. The founder-led brands have tension. Taste. A reason to exist beyond the trends. The filler-led ones? They’re perfectly optimised and entirely forgettable.
That’s what we mean when we say AI is flattening the edges. It’s not destroying design. It’s destroying the need to wrestle with hard questions. To find the sharp angle. To sit in the ambiguity of “why us?” until something real emerges.
AI makes brand-building feel done too soon.
Here’s what we use AI for
We’re not here to shame the tools. We’re just here to draw the line.
- We use AI to consolidate founder voice notes into clear strategy threads.
- We use AI to explore unexpected videography and animation.
- We use ChatGPT to map out what everyone else is already saying.
But we don’t use AI to:
- Name a brand
- Define its tone of voice
- Write its origin story
- Decide which brand elements get designed and why
Because those things are sacred. They don’t come from data. They come from decisions. From doubt. From lived experience and latent tension and the weird way a founder phrases one specific thing in a 6am email.
AI can’t catch that. Because it’s not trained to. And even if it could, it wouldn’t know what it meant to us mere humans.
What we’re seeing now is this mass convergence: of visuals, of copy, of mission statements. Brands that started with potential lose their edge by month three. Not because they’ve failed, but because they’ve been AI’d into oblivion. Polished into sameness. That’s the risk. Not obsolescence. Indistinction.
What we tell our clients
Use AI to make the execution easier. But never let it replace the existential work.
The job of a brand isn’t to look like a brand. It’s to claim space. To express an opinion. To make the founder’s vision legible and magnetic.
If AI helps you do that faster, great. But if it’s the only thing doing the work, your brand won’t last. Because someone else will prompt something similar five minutes later.
What you can do next
Ask yourself:
- Would this brand still resonate if the logo disappeared?
- Could someone else prompt this same story?
- Is this brand specific to me, or just specific to the category?
If the answer feels murky, it’s not a sign you’ve failed. It’s a sign there’s more work to do. And that’s the good part.
Because when you commit to building something real, the AI tools become incredible assets, not crutches.
And when the soul of your brand is sharp enough, clear enough, and rooted deep enough, you’ll know exactly what to automate.
And what should never, ever be touched.