a hidden cheat code for narratives
Last week I wrote that AI doesn’t create meaning. It scales yours. You feel it before you can name it.
You ask AI for help with something that matters. Your strategy. The executive update. Your go-to-market.
The thing you’ve been trying to say about where the company is headed. What comes back with AI is fluent, confident...and completely generic. So you rewrite the prompt. You add context, you feed it examples, you tell it the tone. It comes back smoother, and still not yours. You go again. And again. Twenty minutes in, you aren’t writing anymore, you’re negotiating with a machine that keeps handing you the average and calling it your voice.
It isn’t the model’s fault. You handed it a task, and it reached for the middle, because you never gave it the one thing that doesn’t live in the middle.
The Word Behind the Whole Thing
There’s a word for what you didn’t give it.
It’s a word my wife teases me about, because I bring it up at dinner, at parties, and I’m probably a bit too obsessed about it.
Ontology.
The textbook version: ontology is a branch of philosophy concerned with what exists and how we sort the world into categories. In plain terms, it’s about the labels we put on things, and the assumptions riding inside each label.
My favorite way to show it is with a finger.
Hold up a finger and ask: if this is a number, what is it? One. Now point it up: if this is a direction, what is it? Up. Now wiggle it: if this is a body part, what is it? A finger. One object. Three labels. Each label loads a different world onto the exact same thing.
Put it as a principle: words create worlds. The label you choose decides the world the reader steps into. This isn’t a parlor trick. It’s the foundation under everything I do with narrative, and it’s the step most people never make explicit when working with AI.
The AI Cheat Code
This is where ontology pays off.
Take “platform.” To an LLM, it’s the average of ten thousand SaaS landing pages. To your company it means something specific, and if you never say what, the machine fills in the generic version, and so does everyone who reads what it wrote. Same word, two different worlds.
So ask AI to take an ontological approach, to break down the ontology of your paragraph, and it stops guessing and starts showing you the assumptions buried in your own language. The first time, it feels like peeking behind the Matrix. You watch it reach for the average in real time, and you see exactly where to hand it yours instead.
That’s the whole game. AI fluency is the ability to hand the machine your meaning, your ontology, the problem as only you see it, so what comes back could only have come from you. It was never a software skill. It’s a language skill.
The machine scales whatever you give it: your meaning, or the average.
Why This Is So Easy to Get Wrong
Every time you prompt, you’re asking AI to make a hundred small choices about what your words mean. Left alone, it guesses, and it guesses toward the average.
You don’t see it happen. You just end up three drafts deep, down a blind alley, wondering why nothing lands, never noticing the trap door you walked through two prompts ago when it quietly decided “growth,” or “platform,” or “customer” meant the generic thing instead of your thing.
This is harder than it’s ever been, and not because you’re bad at it.
Everything is moving at once, and the words you reach for keep shifting under you. Getting clear on what you actually mean is the real work. Naming the ontology is how you do that on purpose, instead of leaving it to the machine.
Try This Before Your Next Draft
Five minutes, four steps, and the last one is where it cracks open.
Take something real: your positioning, an executive update, an investor pitch, the thing you keep trying to say but haven’t quite nailed yet.
Write it cold. Ask AI to draft it the way you normally would, and save what it gives you.
Run the cheat code. Paste this: “Take an ontological approach. List the key words in what you just wrote, and for each one, tell me what you assumed it means, and where it could mean something different to someone else.”
Analyze the response. What comes back is likely the average it reached for: your own language with generic meaning filled in. It’s easy to miss that this even happened.
Hand it your ontology. Paste this: “Here is what that word actually means in my world, and the one thing I believe about this that most others overlook or don’t understand. Rewrite it from there.”
Now what comes back probably sounds a bit more like you, because you stopped letting it guess your meaning and gave it the real one. The distance between step 1 and step 4 is the path of fluency.
You didn’t get better at using the tool. You taught it your ontology.
What’s true for you is true for the whole company. A company that never names its own ontology hands every employee, and every agent, the same averaged language, and the whole org ends up sounding like the category, selling its wares at a swap meet.
AI was never going to make you sound like you. It scales the meaning you bring, or it scales the average. Naming what you actually mean, and who you’re becoming as you build, is the one job that stays yours. That’s what this series is about.
It’s also why I’ve spent the last eighteen months building Storied AI: a set of skills inside your own Claude or ChatGPT that make your meaning legible to the rooms that decide, the boardroom, the investor meeting, the buyer’s shortlist, the all-hands. So your work gets funded, bought, and believed in. It’s almost ready, and readers here get the first look.
The good news is you can start today, by hand, with the move above. Over the next few weeks I’ll go deeper: how this works, and how to make it hold across a whole team, not just one chat window.
Michael
P.S. Other words my wife is sick of hearing me say: mouthfeel, micro-climate, and tokenmaxxing.
P.S.S. When you’re ready, here’s three ways I can help:
If you want a second set of eyes on your narrative, reply and tell me more
If you need narrative architecture, I help CEOs & operators do just that. Apply for Q3
If you need a keynote speaker, I do them on a select basis. Let’s talk