When AI first showed up in our workflows, every conversation was about what it could do. Generate logos. Spin up moodboards. Produce twelve variations of a tagline before lunch. And yes, it does all of that. But the longer I work with it, the more I realize the real lesson is in what it can’t do. Or more honestly, what it shouldn’t.
It’s the limits, not the capabilities, that are teaching me the most.
What is AI genuinely good at in brand work?
AI is a pattern recognition machine. It pulls from what already exists and arranges it in new combinations. That is genuinely useful. It is also the exact opposite of what brand design is supposed to do.
Brand design, at its core, is the visual and emotional fingerprint of a business. It’s what a person feels when they see your logo on a coffee mug. It’s the reason they remember you weeks after the meeting. That feeling didn’t come from a pattern. It came from a point of view.
Here’s what we’ve found at Beacon. AI is fantastic at the variations stage. Once you’ve made the original creative call, AI will help you scale it across formats, sizes, and channels faster than any team I’ve ever worked with. That’s a real, durable benefit. We use it almost every day.
What it cannot do is make the original call.
“AI is fantastic at the variations stage. What it cannot do is make the original call.”
What does the data actually say about AI in marketing?
A recent Anthropic research paper by Maxim Massenkoff and Peter McCrory measured how AI is actually being used in the labor market, not just where it could theoretically be used. Marketing specialists rank in the top five most AI-exposed occupations, with about 65% of marketing tasks observed in real AI use today.
But here’s what I keep thinking about. That 65% is where commodity work lives. The variations. The repetition. The mechanical outputs. It is not where original brand identity lives.
When AI generates a brand for you, it isn’t choosing. It is averaging. And an average, by definition, looks like everyone else. In a category like behavioral health, where every brand is fighting to feel trustworthy, looking like everyone else is the opposite of the goal. It’s the exact thing that makes a person scroll past.
“When AI generates a brand for you, it isn’t choosing. It is averaging. And an average, by definition, looks like everyone else.”
Where does AI keep breaking on brand projects?
I want to be specific about this, because “AI has limits” is too vague to be useful. After three years of running AI-assisted brand work through real client projects, here are the breakage points I see most often.
The first is conviction. AI will give you ten options. It will not tell you which one is right. It can’t, because it doesn’t have skin in the game. It hasn’t sat across from your ideal client, watched them flinch at the wrong word, watched them lean in at the right one. The pick, the actual decision about what your brand is going to be, is a human act of conviction. AI can prepare the table. It cannot serve the meal.
The second is context. AI doesn’t know that the founder of the practice lost her sister to addiction and built the company in her memory. It doesn’t know the local market has been burned by a chain that made a lot of promises and disappeared. It doesn’t know that the lead therapist’s calm voice is the actual reason patients refer their friends. All of that is the brand. None of it is in the training data.
The third is taste. Taste is a slow-built thing. It comes from a thousand small decisions made over years, watching what landed and what didn’t, what aged well and what got tired. AI can imitate the surface of taste. It can’t carry the judgment underneath it. Research from the Nielsen Norman Group on brand trust shows users decide whether they trust a brand within seconds, and most of what they’re reading is signals AI doesn’t know how to fake.
When I see a brand that feels indistinguishable from every other brand in its category, I usually find one of these three breakage points behind it.
Why does this matter more in behavioral health?
In behavioral health, your brand is doing more than marketing work. It is doing trust work.
A potential patient who lands on your website is often in one of the hardest moments of their life. They’re scared. They’ve maybe tried before and been disappointed. They are scanning everything you put in front of them for one simple signal. Are these people real, and can I trust them with something fragile?
A brand assembled from the average of what’s out there will not pass that test. It might be polished. It might even be pretty. But “polished and pretty” is not what someone in crisis is filtering for. They are filtering for human. They are filtering for specific. They are filtering for “someone built this on purpose, with me in mind.”
This is where AI’s averaging tendency does the most damage. The exact moment a brand needs to feel most distinctly human is the moment AI is least equipped to deliver. Not because the technology is bad, but because the work itself is outside what pattern recognition can produce. The same logic applies to your website design and your content strategy. The patient is filtering every layer for that signal.
“The exact moment a brand needs to feel most distinctly human is the moment AI is least equipped to deliver.”
What does my grandmother’s gallery have to do with AI?
My grandmother, together with a partner, opened the first art gallery in Anchorage, Alaska in 1971. There wasn’t a best practice for that. She wasn’t optimizing for anything. She was building something where there had been nothing.
I think about that often when I’m watching AI do its thing on a brand project. AI can only see what already exists. It can rearrange, recombine, recolor, and reformat. But it cannot stand in front of a blank wall and decide what should be there. That part is still human work, and I’d argue it is becoming more valuable, not less.
The same goes for the harder calls inside a project. Naming. Voice. The visual choice that doesn’t follow the trend report because the trend report is what every competitor is also reading. AI is great at the trend. It is not great at the deliberate departure from it. That gap matters more as more brands route through the same tools. A Harvard Business Review piece on AI and strategic differentiation makes a similar argument: when everyone has access to the same generators, the value moves to the human judgment around them.
I came up in a family that built things where best practices didn’t yet exist. My grandmother in Anchorage. My father starting a hybrid non-profit/for-profit company in the 90s when the state told him it had never been done. The lesson in both cases was the same. The most valuable thing you can build is the thing that isn’t already in the data.
So what is the real lesson?
Here’s what this is really about. Your brand is your business’s voice when you’re not in the room. Whether someone hires you, refers you, or trusts you with their care depends on that voice feeling like a real human point of view. Not the average of every business that scraped the same training data.
AI is an assist. It is not a replacement. And in brand work specifically, the captain matters more than almost anywhere else, because the audience can feel the difference between a brand made by someone who knew exactly what they wanted and a brand generated to fill the space. They might not name it. They will absolutely act on it.
The limits AI keeps showing me aren’t a reason to keep it out of brand work. They’re a reason to know exactly where the human has to stay in. The variations are the easy part now. The conviction, the context, the taste, the original call. That’s the work that didn’t get cheaper. If anything, it got more valuable, because everything around it got faster.
This is why we’re so deliberate about how AI shows up inside our marketing strategy work for clients. Not because we’re afraid of the tools. Because we know which jobs they’re built for, and which jobs they aren’t.
“The variations are the easy part now. The conviction, the context, the taste, the original call. That’s the work that didn’t get cheaper.”
For my fellow founders: where have you felt AI quietly nudging your brand toward sameness? I want to hear about it.
