Most "how to use AI for brand positioning" advice is wrong. Not subtly wrong. Categorically wrong. The advice usually goes: feed the AI your brand info and ask it to suggest positions. The AI dutifully suggests three positions that all sound like positioning consultancy output. The team picks the safest one. The brand ends up with the same kind of positioning every other brand in the category has.
That's not a tool problem. It's a method problem. Here's how to actually use AI for positioning work, after running this hundreds of times on real brands.
STOP ASKING AI WHAT YOUR POSITION SHOULD BE
The AI doesn't know what your position should be. It can't. It doesn't have access to the things that determine the right position: your real customer behavior, your competitive context, your team's actual capacity, the specific moment in the market, your conviction about what you want to be. The AI has access to a synthesis of every "positioning best practices" article ever written, which is exactly the wrong input for arriving at a non-generic position.
Ask the AI to make positions, you'll get the consensus. The consensus is the position that anyone could arrive at, which is by definition not yours.
USE AI FOR THE STAGES WHERE IT'S ACTUALLY GOOD
AI is excellent at three specific stages of positioning work:
Mapping the competitive landscape. Give it the top five competitors in your category and ask what they're all implicitly agreeing on. The AI is good at synthesis. The implicit consensus is the thing nobody is challenging — and the unowned position is usually the inverse of that consensus.
Pressure-testing your hypothesis. Once you have a draft position, ask the AI to take it apart from the perspective of a sharp competitor. The AI is good at finding holes because finding holes is just pattern matching against weakness. This is way more useful than asking it to generate positions.
Generating language variants. Once the position is set, the AI is excellent at producing twenty different ways to say the same thing. The strategic call is yours. The wordsmithing can be partially outsourced.
THE METHOD
Step 1: Map the consensus. Paste in your competitors' homepage copy, taglines, and positioning statements. Ask the AI to identify the implicit assumptions everyone in the category is making. Look for the thing nobody is challenging.
Step 2: Generate inversions. Ask the AI to produce three positions that would only work if you rejected the consensus assumption. These are the candidate positions. They'll feel uncomfortable. That's the signal you're outside the consensus.
Step 3: Pressure-test the candidate. Pick the candidate position that scares you most. Ask the AI to take it apart from a competitor's perspective. Ask what assumptions it rests on. Ask what would have to be true for it to work. The position should survive at least most of the interrogation.
Step 4: Sharpen the language. Once the position is set, ask the AI for twenty ways to express it. Pick the version that's the most specific and the least like marketing copy.
Step 5: Make the call yourself. Don't ask the AI to choose. The choice requires conviction the model doesn't have. The AI's job ended at step 4.
WHY THIS WORKS BETTER
This method uses the AI for what it's good at (synthesis, pattern matching, language variation) and reserves what it's bad at (strategic conviction, judgment under uncertainty, the willingness to be wrong) for the human. The output is a position that came from your thinking, not the model's averaging.
Most teams skip this method because it's slower than asking the AI for positioning. Yes. It's slower. The output is also actually different from what your competitors will arrive at. That's the whole point.