your ai already has a theory of you
Before I get into it: a lot is taking shape at Storied right now.
A new website next month, some revamped offerings, and a Storied AI plugin that you can use in Claude or ChatGPT, along with an enterprise solution for your language system. For the last 18 months I’ve been building a narrative protocol, drawn from decades of my work. I can’t wait to put it in your hands.
This is the first piece in a short series that runs underneath all of it.
It starts with something a lot of us are feeling and not saying out loud: AI made the work faster, but somehow left us less sure of who we’re becoming in the process.
Your AI Already Has a Theory of You
You never approved it. It just started filling in the blanks.
It built that theory the only way it could. Every time you prompt it, the model reaches for three sources: (1) the prompt in front of it, (2) the internet’s idea of you, and (3) what it infers from your past behavior. Every prompt, edit, and upload is exhaust, and the model reads all of it.
Those three inputs share one blind spot. Every one is assembled from where you’ve already been, not where you want to go.
It shows up when you ask it to write your positioning. It comes back fast and fluent: “We help modern teams do their best work, all in one place.” It sounds pretty good on the surface. But is it ownable? Defensible? Durable? It could be any company in your category, because AI is trained on the middle.
By the third pass you’re rewriting it by hand. The reflex is to call it a model problem. It isn’t.
When your company has never written down what it believes, the model has no center to hold to, so it averages what it finds.
AI doesn’t create meaning. It scales whatever meaning it inherits.
A Prompt Is a Set of Directions
So is an agentic workflow, and every brief you hand your AI agents.
ead your own and you’ll find outcomes every time: the format, the audience, the job to be done. The model learns everything about what you want produced and nothing about who you’re becoming by producing it.
An output is easy to copy. Copy the file and you get the file.
You don’t get the company that wrote it.
Becoming Needs Language First
Who you want to become matters more than the next milestone.
Because it decides what’s even worth reaching for. AI is built for achievements and milestones. Hand it an outcome and it’ll draft the launch and compress the roadmap. What it can’t do is tell you who you want to become. That hasn’t happened yet, so it’s left no exhaust to read.
Here’s the punchline. You can’t build toward what you can’t yet put into words. Becoming, even half-formed, has to be written down in language specific enough to hold something still taking shape. That language is both the vector and the container.
I once worked with a team inside a large company that everyone treated as the “cop on the beat” that slowed every launch down. That wasn’t who they were becoming. They were turning into the group that helped the business move faster and safer at once. But nobody had the words for the shift, so the old reputation kept winning. Once they wrote the new one down, in plain language their partners could repeat, the story started to travel ahead of them.
This is usually when someone tells me they’re too early to work on their narrative, that they still need to figure out what they’re all about. It’s actually the opposite.
It’s in those moments you don’t feel ready that you need it most.
Identity Needs a Container
So before you tune another prompt, do the harder thing first.
Capture who you want to become, and write it down in language only you could have written.
The easiest way in isn’t to write. It’s to talk out loud, stream of consciousness.
I use Wispr Flow for voice-to-text; you can also use the native voice-to-text in ChatGPT or Claude. The point is to think out loud, contradictions and all, and let the transcript be the raw material and your authentic voice. You don’t need a perfect prompt.
You need honest answers to questions most teams never sit with:
The noble purpose underneath the work
Your contrarian view about where the industry is headed
What’s missing from your category’s conversation that matters most
Where difficult dilemmas are quietly pulling the company’s identity apart
Answer those and you’ve started something the model can’t.
As Lisa Cron puts it, you have to go from generics to specifics.
The more specific the story, the more universal the message. If you don’t get specific enough, your voice and personality and humanity will never shine through. Those choices are how meaning gets made. And AI can’t make meaning.
Whose Language Wins
As the ground shifts, everyone is quietly asking whether any of it still makes sense.
That’s an identity question, and a narrative is what answers it. A narrative is the throughline that shows people who you’re becoming as an org and where they fit in, so the shift reads as direction instead of drift. This is an existential moment.
Naming who you’re becoming is the way through.
That’s what this series is about.
Over the coming weeks I’ll get specific about how you build it: how to name who you’re becoming, and put it in language only you could have written, so your AI scales that instead of the internet’s idea of you.
The machine will scale your language either way. The only question is the source: drawing from where you’ve already been, or who you’re seeking to become.