“Abracadabra. I create as I speak.”— Ancient Aramaic
For twenty-five years, language was my laboratory.
Not theoretically. Not academically. In the most practical sense available — in therapy rooms, in training halls, in the moments when the right word, asked in the right way, to the right part of a person, changed something that had been unchanged for years.
I understood how questions open perception. How framing shapes interpretation. How the structure of a sentence can bypass resistance, or invite it. How meaning is not transmitted — it is constructed, in the space between the speaker and the listener, by forces that are mostly invisible and almost entirely linguistic.
Language was my craft. My medicine. My technology.
And then AI arrived.And I used it like Google.The first chapter — information without intelligence
I asked for summaries. For ideas. For answers. For the kind of information retrieval that search engines had been providing for decades, now delivered in conversational form.
And honestly — I was not impressed.
Useful. Fast. Occasionally surprising. But not revolutionary. Not the thing that the people around me — my partner deep inside the world of developers, researchers, inventors, people building systems I could barely name at the time — were so animated about.
I found myself, somewhat accidentally, inside what I came to think of as a laboratory of the future. People speaking in code. In architectures, agents, automations, and models. People for whom the technology was not a tool but a territory — something to be explored, mapped, built within.
And my partner kept saying, with a patience I did not fully deserve:
“Use your skills. This is the future of AI.”
I laughed.
There was an irony in my laughter that I didn’t fully appreciate at the time.
In 1992, I had studied programming and systems analysis — had touched the early language of machines, had learned to think in logic and structure and conditional sequences. Then my life took me somewhere else. Into human code. Into the patterns beneath behavior, the beliefs beneath emotion, the linguistic architecture beneath everything a person does and doesn’t do.
For decades, I moved from computer logic into human transformation.
And now, standing in this laboratory of geeks, I could feel the two worlds circling each other again.
I kept using AI like Google.Until the day I watched them build.The moment everything changed
Not writing prompts. Building systems. Creating workflows. Making AI do things that required not just questions but architecture — structured sequences of instruction that directed attention, constrained reasoning, assigned identity, specified context, shaped the entire cognitive process of the interaction.
And I saw the problem immediately.
Because I am a linguist. A behavioral analyst. A person who has spent a career watching what happens when language is imprecise — when the question is too vague, the frame too general, the intention too unclear.
The prompts were full of generalizations.
The questions asked for intelligence without providing the linguistic structure for thinking.
And the AI — obediently, inevitably — was reflecting the quality of the input back as the quality of the output.
General question. General answer.
Vague frame. Vague result.
Weak intention. Directionless response.
This was not an AI problem.
This was a language problem.
And language problems are what I do.
The experiment
We started to play.
We chunked down — moving from the abstract to the specific, from the general to the precise. We tested different frames, different metaphors, different levels of abstraction. We assigned identities, belief systems, perceptual positions, reasoning processes. We constructed questions the way I had spent twenty-five years constructing questions for human beings — with intention, with context, with a clear understanding of what we were trying to open.
And the results changed.
Not incrementally. Completely.
The same AI. The same underlying model. A different linguistic construction — and a different perception, a different reasoning path, a different quality of intelligence emerging from the interaction.
That was the moment I understood.
Prompting is not typing.
Prompting is not asking.
Prompting is not giving commands to a machine.
Prompting is linguistic engineering.
It is the design of attention.
And when attention changes, perception changes. When perception changes, reasoning changes. When reasoning changes, what becomes possible changes.
The spell
The ancient word abracadabra comes from Aramaic.
Avra kadavra. I create as I speak.
The oldest articulation of what every serious practitioner of language has always known — that words are not merely descriptions of reality. They are interventions in it. Structured sequences of linguistic input that change what the receiver attends to, how they interpret what they attend to, and therefore what they do.
In Neuro Linguistic programming terms, we call this the linguistic frame. In cognitive science, it is the attentional set. In rhetoric, it is the construction of the possible. In therapy, it is the question that opens a door the client didn’t know existed.
In AI, it turns out, it is the prompt.
A spell. A structured sequence of words that directs a mind — biological or artificial — toward a specific way of seeing, reasoning, and acting.
The principles are identical. The medium is different.
What I built — and what it revealed
Five hours after this understanding arrived, I had built my first AI agent.
From zero practical AI-building experience. No coding. No technical background beyond what I had touched in 1992.
Five hours earlier, I was not building agents. Five hours later I had created something that could think from a role, follow a reasoning process, organize information, respond to a specific need, and produce output that reflected genuine structured intelligence.
Not because I had learned to code.
Because I already knew how to construct linguistic architecture for minds.
I had been doing it for twenty-five years.
With humans.
And in that moment — standing in my own version of the laboratory, looking at what had been built in an afternoon — I saw a door.
A door of perception.
The future of AI, I realized, does not belong only to those who understand machines.
It belongs equally to those who understand minds.
The trainers. The therapists. The coaches. The facilitators. The researchers. The communicators. The people who have spent careers understanding how language directs attention, how questions open perception, how meaning is constructed, how beliefs shape interpretation, how structure creates behavior.
We are not late to this technology.
We may be exactly on time.
The new field
What I am pointing toward is not prompting as a productivity skill.
It is something I would call Intelligence Engineering — the design of shared cognition between human and artificial intelligence.
The capacity to construct the linguistic conditions under which AI becomes not a tool that retrieves but a partner that reasons. Not a search engine but a thinking system. Not a generator of content but a collaborator in the construction of something genuinely new.
This requires everything that behavioral science, linguistics, and human transformation have been developing for decades.
Rapport — because the quality of the relationship shapes the quality of what is possible within it, even when one party is artificial.
Context — because intelligence without context is pattern-matching without understanding.
Precision — because vagueness in the input produces vagueness in the output, in any mind, biological or artificial.
Framing — because the frame determines what can be seen within it, and what remains invisible outside it.
Belief systems — because the identity and values assigned to an AI system shape its reasoning the same way identity and values shape human reasoning.
Chunking — because the level of abstraction at which a question is asked determines the level of abstraction at which it is answered.
These are not AI concepts. They are human ones. They emerged from decades of careful observation of how human minds construct meaning and produce behavior.
And they transfer — with remarkable precision — to the design of artificial intelligence interactions.
What this means for the behavioral edge
I want to connect this explicitly to the territory this newsletter has been mapping.
Every issue has been exploring the same fundamental question from different angles — how do human beings construct their experience of reality, and what becomes possible when they do it more consciously?
The metacognition. The somatic awareness. The shadow. The map that is not the territory. The triggers. The masks. The inherited definitions of right. The freeze. The betrayal. The promotion that completes the disappearance.
All of it is about the gap between the reality we construct — through language, through belief, through the frames we have inherited and the ones we have chosen — and the reality that might be available if we constructed it more deliberately.
AI is the newest frontier of exactly this question.
Because AI constructs its reality — its response, its reasoning, its output — from the linguistic input it receives.
Which means that working with AI is, in the most precise sense, an exercise in conscious reality construction.
The quality of the spell determines the quality of what is conjured.
And the people who understand spells — who have spent careers studying how language creates reality — are exactly the people this new world needs.
A final thought
I was laughing at the beginning.
In the laboratory of geeks, at the edge of a territory I didn’t yet understand, genuinely uncertain whether what my partner was pointing at was as significant as she believed.
I am not laughing now.
Not because the technology has impressed me into reverence. But because I walked through the door — and on the other side, I found something I already knew.
The principles I had been teaching for twenty-five years were waiting for me in a new medium. The linguistic intelligence I had developed for human minds transferred — immediately, powerfully, almost entirely — to the design of artificial ones.
Abracadabra.
I create as I speak.
We always have.
Now we are doing it with machines.
And the spell — structured, intentional, precisely constructed — is the most important skill in the room.
Nikolas Fragkias
