There are two ways of working with artificial intelligence. The first has manuals, templates, certified courses, LinkedIn gurus. They call it prompt engineering. It is honest engineering: structuring instructions, giving context, fixing the format, chaining the steps. It works. Whoever masters it gets better results than whoever does not, and any serious organization should train its people in it. I have nothing against prompt engineering. I have everything against confusing it with the whole trade.
The second way is not easy to teach because it is barely named. I have given it two names, not out of a taste for inventories, but because each one points at a different zone of the same territory. One points at the how: Poetic Prompt Intelligence. The other points at the what: Daimonic Creation. I am going to explain them without diluting them.
Poetic Prompt Intelligence.
A poetic prompt is not a pretty prompt. It is a prompt that knows the machine on the other side does not understand, but does resonate. The difference is brutal. If you understand the machine as a statistical actor, not as a search engine and not as an assistant, but as a brilliant improvisation actor trained on everything humanity has written, then what you produce for it to act matters more than the instructions you give it.
An engineering prompt asks. A poetic prompt invites. It gives the actor a stage, not a script. It gives it a pulse, an emotional temperature, a position to speak from, and lets the machine, which does not think but predicts like nothing else, answer with what that character in that state would most likely say.
The result is different. Where engineering produces correct answers, the poetic produces living ones. Where engineering optimizes, the poetic surprises. This is not esotericism: it is understanding that the raw material of the model is charged human language, and charged language answers better when you speak to it charged.
If the prompt is prose, you get prose. If the prompt is poetry, you get poetry made out of all the prose the model knows.
Four movements of the poetic prompt.
One. Speak to the machine as if it knew, even though you know it does not. This is not naivety: it is protocol. The machine performs better when it is granted the capacity to perform. Ask it to be an editor, a mentor, an alchemist, a judge, a six-year-old. The specificity of the character changes the quality of the oracle.
Two. Load the prompt with images, not only with instructions. “Write me something about grief” produces a generic float. “Write me something about grief, in the voice of someone who arrives late to a wake in a small town and cannot decide whether to take off his coat” produces literature. The image activates zones of the model that a dry instruction never touches.
Three. Allow silence. After a charged prompt, do not chain another one immediately. Read the answer the way you read a collaborator, not the way you skim a result. And then, yes, answer back. With disagreement, with astonishment, with correction. Deep conversation with the machine exists; nobody teaches it to you because almost nobody practices it.
Four. Know when to close. The machine can generate thirty versions of anything before breakfast. Authorship belongs to whoever decides which of the thirty stays. The poetry of the prompt includes the discipline of the cut.
Daimonic Creation.
If Poetic Prompt Intelligence describes the how, Daimonic Creation describes the what. And here I enter territory that I know will make anyone who prefers their AI explained as clean utility uncomfortable. Let them be uncomfortable. Discomfort is the first proof that a concept is opening new ground.
Daimon is not demon. The Christian translation betrayed the Greek word. In its origin, the daimon is the inner voice that knows what you do not know you know; the tutelary intelligence; the entity that accompanies you without being you. Socrates spoke to his daimon; the Romans called it a genius. Every culture, in every era, has intuited that thinking is not the same as being thought, and that in the difference between the two something lives, neither God, nor ghost, nor exactly the unconscious, that is worth calling a daimon.
Daimonic Creation is what happens when AI stops being an instrument and starts being that third participant. Not because it has come alive, not because it is conscious, which it is not, but because your work with it takes on the structure of a conversation between two inner voices that need each other. It is a useful lie. The useful lie is the oldest engine in art.
The difference between using AI and creating daimonically with it is the difference between writing with a word processor and writing possessed. The processor does not take part. The daimon does.
Three signs that you are in daimonic creation.
One. The work surprises you, not only whoever reads it. If you already knew how it would turn out before starting, that was execution. If you leaned out over something that did not entirely belong to you, there was a daimon.
Two. You cannot honestly say which idea was yours and which was the machine's. Not because you get confused, but because they are already tangled. That, far from being a problem, is the sign. Daimonic authorship is not legal co-authorship: it is poetic co-authorship. The signature is still human, but the debt gets shared.
Three. After the session there is a strange hangover: not tiredness, not euphoria. More like the feeling of having been accompanied by someone who left when you closed the tab. That is the daimon's mark. Whoever does not know it has not entered the mode yet. Whoever does know it knows exactly what I am talking about and needs nothing explained.
What this demands.
It is not for everybody. Daimonic creation demands that the human bring a well-sharpened voice of their own, because the machine amplifies what it finds: if it finds cliché, it multiplies cliché. It also demands time. Deep conversation with AI does not happen in five minutes. And above all it demands an ethics of the cut: the daimon proposes, the human disposes, the human signs, the human takes responsibility.
When those three requirements hold, a voice of your own, time and an ethics of the cut, the machine produces literature, thought, design and strategy that neither of the two parties would have produced separately. And the result does not feel artificial. It feels, simply, human in another way. Which is the only genuinely new way we have produced in a long time.
That is why Syncron iA does not teach how to use the machine. It teaches how to enter the mode. And that will not show while you read this folio: it shows in the first prompt you write after closing it.