AI has become genuinely useful in my day-to-day brand work. It drafts faster than I can, analyzes competitor positioning, generates visual directions, pressure-tests messaging. I'd be slower without it.
But when a client asks me should our brand say X or Y? — I don't ask an LLM. I still do that work myself.
Here's why the line hasn't moved, and probably won't.
Strategy is a claim, not a summary
A brand strategy is fundamentally a commitment. You're saying: we are this thing, we serve these people, we do not serve these other people. The strategy's value comes from the exclusion.
LLMs, at their core, are averaging machines. They've read everything and produce the weighted-average acceptable answer. Averaging is the opposite of strategic commitment. A strategy that's the average of all strategies is definitionally not a strategy.
Ask an LLM "what should our positioning be?" and you get something reasonable that no one will remember — because reasonable is what everyone else already has.
Taste is a bet, not a calculation
Brand work requires aesthetic and tonal judgment — which typeface feels right, which color evokes the correct emotion, which tone lands with this audience.
LLMs are improving at imitating taste. They're not getting closer to having taste. Taste is an investment in one direction at the cost of another, made by someone who will live with the consequence. Nothing about current AI architecture rewards making expensive commitments.
A designer who signs off on a color has skin in the game. The client fires them if it's wrong. An LLM that suggests a color has no stake. Its suggestion will be tepid because it's socialized into tepidness — training data penalizes strong positions.
Context that lives in conversations
A lot of brand strategy is encoded in conversations that never get written down:
- The founder's offhand comment about what they hate
- The way the operations lead described a customer at the kitchen table
- The investor meeting that clarified which segment they won't serve
- The co-founder's visceral reaction to a competitor's deck
This context isn't in any document the AI can read. It lives in the people running the business. Good strategy extracts and codifies this context. LLMs can help write it up. They can't surface it.
Where AI is great in the strategy workflow
To be clear — I'm not anti-AI in strategy work. Here's what I use it for constantly:
- Competitive synthesis. Summarizing 40 competitor sites into patterns I can react to.
- Draft pressure-testing. "Here's our positioning statement. What are three obvious rebuttals a customer might have?"
- Message variation. Generating 20 riffs on a tagline to find the one that feels right.
- Research interviews. Transcribing, clustering, tagging. LLMs are excellent at this.
In all these, the AI is augmenting my decisions, not making them. The decisions remain mine.
The honest version
If a client is paying for our brand strategy engagement and asks if I use AI — the answer is: yes, constantly, as a research assistant. But the strategic calls are mine, named, and signed. That's what they're paying for.
Anyone telling you their AI is the strategy — check who's signing the deck. If it's the LLM, run.