You have tried to make AI sound like you. You gave it your bio. It still sounded generic. So you gave it more, your background, your approach, your specialties, maybe a personality test result you took once. And it got more detailed and no more you. At some point you concluded that AI just cannot capture a real voice, or at least not yours, and you went back to writing everything by hand or shipping content that quietly embarrasses you.
The conclusion was wrong, but it was reasonable, because you were doing the one thing that was never going to work, and nobody told you it was the wrong thing. The reason AI cannot sound like you is not the prompt, not the model, and not some limit on what these tools can do. It is that you have been feeding the AI the wrong kind of input. This is the cognitive context problem, and once you see it, the fix is obvious.
Why can't AI sound like me even when I give it lots of information about myself?
AI cannot sound like you no matter how much information you give it, because information is the wrong kind of input: facts about you are things the AI reads, and reading facts does not change how it writes. What changes how it writes is instruction it runs.
This is the whole misunderstanding, so slow down on it. There are two fundamentally different kinds of thing you can hand an AI, and they do different jobs.
The first kind is information. Your bio, your credentials, your list of specialties, your personality-test type. The AI reads this. It is a fact to be consulted. And here is the part that surprises everyone: reading facts about you barely changes how the AI writes, because facts describe who you are, not how you produce. You can tell a tool everything true about yourself and it will still write in the generic average voice, because you told it what to know, not how to write.
The second kind is instruction. Rules for how you actually write and see. "State the specific struggle before offering any reassurance. Use short, direct sentences. Never use the word journey. Never soften a hard truth into a platitude." The AI runs this. It is not a fact to consult, it is a frame to operate inside. And it changes the output completely, because it governs how the tool produces every sentence, not just what it knows.
The reason your effort plateaued is that you kept adding the first kind and never added the second. More information, read and ignored, when what you needed was instruction, run and obeyed. This is not a matter of quantity. It is a matter of kind, and no amount of the wrong kind adds up to the right one. (This is why more facts about yourself never fixed the generic output.)
The difference between reading and running
The distinction underneath the whole problem is read versus run: a fact is read and consulted, an instruction is run and executed, and only the thing that gets run changes how the output actually comes out.
An analogy makes it concrete. Imagine two ways to help someone cook like you. The first is to hand them your biography, where you grew up, what foods you love, your favorite restaurants. They read it, they know a lot about you, and they still cannot cook a single one of your dishes, because knowing about you is not knowing how you cook. The second is to hand them your actual method, the steps, the rules, the things you always do and never do. They run it, and the food comes out like yours, because you gave them the process, not the profile.
Facts about you are the biography. They get read and change nothing about the output. Your voice and method are the recipe. They get run and change everything. Every therapist trying to make AI sound like them by feeding it more facts is handing over the biography and wondering why the food still tastes generic. The fix is to hand over the recipe: the executable instructions for how you actually write, think, and see.
What is "cognitive context"?
Cognitive context is the executable model of how you specifically think, decide, and communicate, written as instructions an AI can run rather than facts it can only read, so the AI produces in your shape instead of falling back to the average.
The term names the missing layer. Regular context is the information you give an AI, the facts, the background, the stuff it reads. Cognitive context is the layer above that: not what you know or who you are, but how you operate, encoded as running instructions. It is the difference between telling a tool about yourself and teaching it to reason and write as you. And it is the exact thing missing when a therapist says "I gave it everything about me and it still sounds generic," because everything about them was information, and cognitive context is instruction.
This is why generic AI writes generic therapists. Given only information, or only a topic, the tool has no cognitive context to run, so it falls back to the only thing it has, the statistical average of all therapist writing, which is the invisible middle. It is not that the tool refuses to sound like you. It is that you never gave it the executable model of you it would need to. Supply that, and the same tool that flattened you now holds your shape. (This is the structural reason a practice goes invisible, now at the level of your tools.)
Why does this matter so much for therapists specifically?
This matters most for therapists because your entire value is genuine, specific human understanding, so content produced without your cognitive context does not just underperform, it signals the absence of the very thing you are selling.
A generic marketing email from a software company is a minor problem. Generic content from a therapist is a contradiction of the core offer. The client is choosing you for specific human understanding, and generic content, content that could belong to any therapist, quietly signals that there may be no specific human here at all. So the stakes of the cognitive context problem are unusually high in your field. Getting your voice into your tools is not a nice optimization. It is protecting the one signal your whole practice depends on: that a real, specific person, who understands a real, specific kind of client, is on the other end of the words. (This is the same reason it matters across your whole practice.)
How do you actually give AI your cognitive context?
You give AI your cognitive context by identifying how you specifically think and write, then encoding it as executable instructions, positive and negative, that the tool runs on every piece, which is real work to do well and is the entire difference between generic and yours.
The doable-by-hand version has three moves. Identify your real position and voice: the specific person you serve, the particular way you see them, the concrete markers of how you write. (This starts with your positioning statement.) Encode it as instruction, not information: how-you-write rules the tool runs, not facts it reads. And specify the negatives: precisely what you never say, so the tool has guardrails that hold your shape across long pieces instead of drifting to the middle in every gap you did not cover.
Done carefully, this works, and you can begin it today. Done at depth, so the instructions are complete enough that the tool holds your shape across everything, not just a paragraph, it is a substantial build, and doing it well by hand is genuinely hard, which is exactly the problem BNA72 was made to solve.
What BNA72 actually is
BNA72 is a cognitive context system. It builds the executable model of how you specifically think, decide, and communicate, from one assessment, so your marketing and your AI tools sound like you specifically instead of like the average therapist.
Here is what it actually does, plainly. You take an assessment that maps how you think, decide, and connect. From that, it generates a written system, your position and your voice encoded as instructions an AI can run, including the negatives, the specific things you would never say, so the tool holds your shape across long pieces instead of drifting. It is not a personality test that tells you about yourself, though it will tell you things. It is a persistent instruction set that teaches your AI tools how you operate, so you stop getting generic output and stop re-explaining yourself to every tool from scratch. The recipe, written once, that every tool can then run.
Three tiers, depending on how far you want to take it.
The $197 Foundation is the assessment and your blueprint: the model of your position and voice, the true thing and the language for it. It is where most people start, and on its own it is the highest-impact move you can make on the invisibility problem.
The $997 Complete builds the blueprint into a full brand and marketing toolkit, the whole system your practice runs on, so the cognitive context shows up everywhere, not just in one document.
The $1,997 Platform is the partnership tier, built for group practice owners solving this across a whole roster, so every clinician gets a distinct cognitive context inside a coherent practice frame, which is the group identity problem solved at the level of the tools. (Here is the full group practice version of this.)
Whatever tier fits, the thing you are buying is the same underneath: cognitive context, the executable model of you, so the tools that are becoming part of every therapist's work amplify your voice instead of averaging it away.
See the drift, then decide
Before anything else, find out how much your AI-assisted content has already drifted toward generic, because most therapists cannot see their own drift, and seeing it is what makes the fix feel urgent instead of optional.
The Generic Voice Checklist walks you through ten signs that AI is flattening your voice, in your own content, so you can see exactly where the cognitive context is missing. Start there. It costs nothing and it will show you, concretely, whether your tools are producing you or producing the average.
And if you want the built version, the executable model of your voice that lets your tools hold your shape instead of erasing it, the Foundation assessment is where that begins. Not because you need it to understand the problem. You understand it now. But because doing the encoding well, at the depth where the tool actually holds your voice across everything you produce, is a real build, and this is the system that does it. The choice, as always, is yours. The diagnosis is free, and it was always the same one: generic AI writes generic therapists, because it was reading facts about you when it needed to be running the instructions that are actually you.