There’s one thing I’ve learned in two years of conversations about AI with entrepreneurs, CEOs, and technical leaders. The decision almost never comes down to the numbers.
The same decision-maker who, faced with an important hire, weighs options, compares, asks for data — with AI goes on gut instinct. And doesn’t notice. They think they’re evaluating a technology. They’re reacting to one, which is a different thing.
It happens because AI arrived without a history. The CEO was already a CEO before, and had other CEOs behind them to learn the role from. The technical leader had their seniors. The CFO too. With AI, no. Nobody has had a teacher. We’re the first generation standing in front of it, and everyone is figuring it out alone, drawing on instinct from experiences that may have nothing to do with AI. So, before reasoning even begins, the posture arrives. It isn’t chosen. It emerges when instinct speaks before reason.
I counted them. Seven recurring postures — in those who decide and in those who build the offer. One is mine. One is probably yours. One belongs to your CFO. We’re all using ours, convinced we’re being objective.
This is the complete map. Not to label, but to understand where you’re looking from — and how you see through that posture.
They’re all gut reactions. They arrive fast, warm, before you’ve had time to form an opinion. The most honest and the most blind, in the same gesture. Even the most reasoned one, as you’ll see, can fire on instinct.
The Renaissance is mine. For half my life I loved building software with my own hands, then had to stop to focus more on running the company. With AI I went back to writing code, and the feeling of superpower — collapsing the gap between idea and execution — was something I’d never felt before. It’s the triumph of experience: someone with years of craft behind them has more ideas than ever, and finally the tool to realise them. The potential is enormous, and it’s fuelled by joy. The limit comes when I try to pass the method on to others. What I do is stitched to my own head, my own instinct. It’s not yet a procedure. And until I notice that, I think the others just aren’t getting there. The truth is more uncomfortable: they have no procedure to follow, because I haven’t built one yet.
The Satisfied you recognise by a single phrase: “for what we do, ChatGPT is enough.” It’s said by whoever throws documents in and reads the answer — and the answer passes the test of their eye. It’s not naivety; it’s an honest observation of the result in front of them. Inside this posture there’s real wisdom: knowing when something is good enough for what’s needed. The limit is that the Satisfied never looks underneath. Of the documents they’ve uploaded, how many has the machine actually read? When two contradict each other, which did it prefer? They don’t ask — not because they don’t care, but because as long as the answer seems right, there’s no reason to go deeper. It’s the quietest of the seven. The Satisfied doesn’t know they’re the Satisfied.
The Honestly Disoriented says “I don’t know” out loud, from a position that institutionally rewards the opposite. An ICT manager from a multinational known for its strong push toward innovation called me: “I have two hundred thousand euros of AI budget and absolutely no idea how to use it.” It’s the sanest thing I’d heard about AI in months. Admitting vulnerability before spending it on rushed decisions is the most respectable posture in the group — and the door through which every movement passes. The limit comes after the admission. When you say “I don’t know,” the first instinct is to look for someone who does — and whoever knows is usually whoever answers loudest: the supplier with the complete package, the ready-made roadmap, the quote that closes the case. The clearest answer is almost never the right one. Honesty risks becoming the sponsor of the wrong choice.
The Quietly Disoriented is the twin of the previous one. Same not-knowing, opposite posture. The same disorientation, instead of declaring it, covered with a language of authority. Faced with ground they don’t master — and can’t admit they don’t master — they make the only move that protects them: they delegate. This isn’t new to AI; it’s always happened with every wave of innovation. Those who lead, placed in front of something unfamiliar, push not just the work downward but the decision itself. And that’s where the damage lies: key choices require real study before deciding, and the understanding you need in order to choose can’t be delegated. You delegate execution, not comprehension. Whoever lets go of that has already let go of the wheel.
The Disappointed has already tried, got burned, and shut the door. “It doesn’t work for us.” I watched a mechanical engineering company bring AI into their Technical Office with a significant budget and a university start-up, and end up with that sentence. Under the surface, the technical team thought it was there to help them; management thought it was there to make the sales team independent from them. Two different projects under the same name, never compared. AI hadn’t failed. The design had. But “it doesn’t work” is more dignified than “we never agreed among ourselves,” and it’s final. The Disappointed’s blind spot is turning a single case into a verdict on the technology. The disappointment is legitimate, and it needs to be let pass. The trap is closing the door before understanding which door it was.
The Sovereign, faced with an AI opportunity, doesn’t ask first “what can I gain” — they ask “what can I lose.” A fashion multinational, one of our clients, banned any AI project for eighteen months: the risk of sending data to servers outside their control was unacceptable, and that was enough. You find the same gesture in the developer who says “nobody writes classes the way I do,” in the professional who sees in AI only the erosion of their craft. Inside there’s something precious and clear-eyed: they’ve seen implications others haven’t looked at. Protecting data, know-how, accountability is a real function, not a fixation. The limit is that the Sovereign calls their own stopping prudence — and it isn’t. Real prudence suspends the risk for a moment, looks at the value, and only then brings the risk back in to mitigate it while pursuing the value. Rigid sovereignty never reaches the “after.”
The Process Thinker is the posture of those who think from the start about how to embed the new component into a process to generate organisational value. They see the system before the people, governance before the tool, organisational coherence before the individual prompt. For them, hallucinations aren’t a reason to stop — they’re the point at which to intervene to capture the value. AI isn’t something you give to people: it’s a component you embed in the process, where it’s needed, with clear responsibilities and verifiable outputs. That’s exactly what many organisations need. And here’s the subtlest blind spot of all: the Process Thinker designs elegant systems that then have to be inhabited by real people. People who are Renaissance types, Satisfied, Disoriented, Disappointed, Sovereigns. The perfect process on paper meets the real postures at work — and becomes a process nobody actually inhabits. Their trap isn’t poor design. It’s believing design is enough.
The problem isn’t the posture. It’s the rigidity.
None of these seven is wrong. I’ve seen them in myself, in my colleagues, in the clients I meet every day. Each one is an honest starting point, and each one carries a true core: the joy of the Renaissance, the sufficiency of the Satisfied, the honesty of the Disoriented, the caution of the Sovereign, the lesson of the Disappointed, the systems vision of the Process Thinker. The problem isn’t where you start. It’s staying fixed there.
Four postures, if they calcify, stop you. The Satisfied prevents you from looking underneath. The Quietly Disoriented delegates the one decision that couldn’t be delegated. The Disappointed closes the wrong door. The Sovereign raises the shield and never lowers it again. In all four cases the same thing happens: the posture stops being a starting point and becomes a destination. And a destination, in today’s AI landscape, is a trap. Because the ground is still moving, and you aren’t.
The point of the map isn’t to pin you to a position. It’s to know where you are, so you can move. I started out 100% the Renaissance. Today I’ve migrated toward the Process Thinker. The door that made me move is exactly the one the Quietly Disoriented refuses: admitting where I was looking from.
The mature posture is a blend, not an eighth box
So where does one move? Not toward a new posture. The “right” one doesn’t exist as a point on the map. It’s a calibration of all seven.
There’s the backbone of the Process Thinker, who sees AI inside the system rather than scattered across seven people using seven different tools. There’s the right dose of Sovereign, who protects what genuinely matters — data, know-how, accountability — and then acts, rather than just protecting. There’s the wisdom of the Satisfied, who knows when to stop: when something works for what’s needed, good enough and useful is the goal, and pushing for perfection is just a tax. There’s the honesty of the Disoriented — the “I don’t know” said out loud that keeps the question open, and the refusal to delegate the understanding the decision requires. There’s the lesson of the Disappointed, who learned on their own skin that what almost always fails is the design, not the tool. And you need a touch of the Renaissance: the wonder, the thrill of watching the distance between idea and realisation collapse. That’s the fuel. Without it, the system is correct — but dead.
The postures that calcify stop you. Calibrated, they complete you. The wall and the gift are the same thing, seen from two different distances.
The first question to ask yourself about AI isn’t “what AI do we need.” It’s “where am I looking at it from.” And right after comes a second one, worth even more: “where am I moving toward.”
The posture you can’t see decides for you. The one you can see, you choose.