I spend a lot of my time between two worlds that aren't supposed to understand each other — a boardroom in Vienna in the morning, a workshop somewhere in the Gulf that same evening — and the thing I keep noticing is how quickly the old advantage is draining out of both. For most of my life, knowing things was the edge. Whoever had read more, analysed more, seen more got the seat and the fee. That edge is going, faster than most people around those tables want to admit, and I don't think we've been honest yet about what it means.
Every generation watches something scarce turn abundant, and the people who built their advantage on the scarce thing are usually the last to see it go. Reading and writing were once a profession in themselves — a scribe, a notary, a person of consequence — until literacy spread and the advantage quietly moved to what you did with the words. Doing arithmetic was once a job you could hold; there were rooms full of people called "computers" who worked sums by hand, until machines made arithmetic free and the value moved to knowing which sums were worth running at all.
We're living through the same shift with intelligence itself, and I mean that plainly, not as a figure of speech. Drafting, research, analysis, translation, code, synthesis — the cognitive work a whole class of careers was built on is getting cheaper by the month. The cost of running the kind of reasoning that sat behind a serious analyst's desk a couple of years ago has fallen more than 280-fold in about eighteen months, and today anyone with a laptop and twenty euros a month has drafting and analysis that used to take a room full of well-paid people. That's already the world we're in.
The conclusion I can't talk myself out of is that if everyone has the same models, the same tools, the same answers, then knowing things is no longer a moat. The expert whose value was having read more than you is now up against a machine that has read everything. So the question I keep coming back to isn't how to stay smarter than the machine — that race is quietly over, and honestly it's fine — it's what stays expensive once intelligence gets cheap.
What stays expensive
When I look closely at what these models actually hand me, I notice what they leave out, and it's almost always the part that mattered.
Trust, first of all. A model can draft the contract, but it can't be the reason two people sign it. Trust gets built slowly, in rooms, over years, through promises kept when it was inconvenient and losses absorbed instead of passed on, and that can't be downloaded or copied — at least not cheaply, and not credibly. Every deal I'm proud of was really closed on trust; the paperwork only wrote it down.
Then judgment, which isn't the same thing as intelligence at all. Intelligence answers the question in front of it. Judgment is knowing which question is worth asking, and which impressive-looking answer to ignore because it doesn't fit this founder, this family business, this market, this year. You get it by carrying the consequences yourself — for me that meant being wrong with my own money on the table, more than once.
And then the whole human layer we're strangely embarrassed to name in a business context. Empathy, more than anything — really reading the person across from you, knowing what a silence means in one room and what a handshake means in another. I speak a few languages, but this is more demanding than language: machines are getting good at translating words, they don't translate worlds, and they don't feel the room. Reliability, loyalty, the willingness to stay when it stops being convenient, an openness that lets you keep learning and unlearning as the ground moves — the same qualities that have always made a person worth working with are quietly becoming the scarce ones, and the hardest to fake.
Taste belongs here too. When anyone can generate a hundred versions of anything, the value stops being in the producing and moves to the choosing — knowing which of the hundred is right, and having the nerve to kill the other ninety-nine.
And then there's the plainest thing of all, the one I spend most of my days on: actually getting it done. Making a business work. Closing the deal. A model can prepare everything around it, but it can't sit across from someone and get them to commit, and it can't carry a project over the line when it starts to wobble halfway through. The doing, the finishing, the following-through — that still falls to people.
The people in between
Looking at all of it together, what strikes me is that almost none of it is knowledge. It lives between people — whether they understand each other, whether they trust each other, and who's actually going to take the thing on and carry it through. Because in the end someone still has to be the one who does it: who puts their name to it, makes it happen, and stays with it until it works — or answers for it when it doesn't. A model can hand you the analysis, but it won't sit in that seat. At least not yet — people are already talking about insurance for AI systems, and maybe some of that gets shared out over time. For now, the person who'll stand behind an outcome and actually push it through to success is the scarce one, and more often than not it's the same person who can stand between worlds that don't otherwise connect.
A model can hand you the analysis, but it won't sit in that seat.
When intelligence was scarce, the economy paid for vertical depth: the specialist who knew one thing better than anyone. Universities and consultancies and corporate ladders were all built to produce and rank that depth. Now depth is something you can rent by the hour. What you can't rent is the horizontal — the person who connects the deep silos to each other and to the real world and is prepared to be accountable for what happens when they do. Those people, call them bridges, are the real asset of the next economy.
I've watched a few kinds up close. There's the bridge between cultures — a company in Austria that needs to build, source or produce in the Gulf, or in Asia, or in China, where the culture is different and the whole basis for trust is different, right down to what a promise is taken to mean. More often than not the two sides can't really work together until someone they both trust, and who genuinely understands both, stands in between — not translating so much as being at home on either side. There's the bridge between the technology and the industry — the breakthroughs happen in the model labs from San Francisco to Paris to Hangzhou, but the money will be made where those labs have never set foot: in manufacturing and logistics, in retail and construction, in law firms and tax offices and the family businesses that quietly run half the economy. The gap between what AI can do and what a 200-person company actually does all day is one of the largest and least crowded opportunities of this decade, and I don't think smarter models close it; people who can stand on the shop floor and in the codebase at the same time close it. And there's the bridge between an idea and a working company — ideas were never the scarce part, and now even writing them up is free, so what's left hard is the unglamorous middle: the first hire, the first contract, the first crisis. Building a company is mostly connecting a possibility to a payroll and staying answerable the whole way down.
One thing I'd add, because it's the part people skip. Being a bridge isn't a position you claim once and keep. The easy connections get crowded and commoditised the moment they're obvious — the same fate as everything else — so the ones that hold are the hard ones, and the real skill isn't standing on any single bridge but being able to keep building the next hard one before the easy version of the last gets crowded and copied. That takes exactly the things you can't fake at speed: trust, a real understanding of both worlds, and a willingness to unlearn what worked last time. A middleman wants the gap to stay open so he can keep charging for it; a bridge only earns the next one by actually closing the last, and over time the market tells them apart.
The real skill isn't standing on any single bridge, but building the next hard one before the easy version gets crowded and copied.
Notes from the field
None of this is theory for me, so let me say where it comes from. I run companies between Vienna and the Middle East. I've built engineering teams in places Western firms find difficult and sold their work to clients Western startups find boring. I've sat at the table for AI products in industries that had never employed a single programmer, and I co-founded a fine-jewellery brand where the whole value is in the one thing software can't touch — the object in your hand and the meaning you attach to it.
A few things I've learned, all of them the slow way. The trust deficit is often the business itself: wherever two worlds want to work together but can't quite trust each other — or, just as often, simply don't understand each other, which is its own kind of hell and kills more deals than price ever does — there's a company waiting to be built, and the harder that gap is to cross, the more defensible it is once you've crossed it. Small and fast beats big and slow more often than people expect — I keep watching a handful of people with judgment and good tools do what used to take a whole department, not because headcount stopped mattering but because the real constraint was always decision speed, and decision speed comes down to trust and judgment, which is to say, to people. And craft turns out to be the counterweight to all of it: the more the world fills with generated everything, the more value runs back to what's visibly human — the handmade piece, the personal commitment, the founder who actually shows up. I didn't start a jewellery brand despite working in AI. I started it because of what AI is doing to everything else.
So what do you actually do
If any of this is right, the practical version is simpler than it sounds. Stop competing on what you know, because whatever you know the machine knows too, or will shortly, and compete instead on the things it can't be — relationships, judgment, presence, following through. Use the free intelligence harder than anyone rather than less, so that the machines do everything machines can do and your own hours go only into what stays expensive. Collect worlds, not just skills, because every field or culture or craft you're genuinely native to multiplies the value of all the others, and those intersections can't really be studied, only lived. And put your name on the work: in a world filling up with anonymous output, a real track record is one of the few things that stays expensive to fake and impossible to borrow.
Where this ends
I don't think free intelligence makes people worth less. I think it's quietly ending a misunderstanding we've carried for a century — that our value lived in what we knew. It never really did. It lived in the trust between two people who shake hands and mean it, in the empathy to understand who is actually across the table, in the judgment to choose a path and answer for it, in actually getting the thing done, and in the openness to keep learning and unlearning as the ground moves under all of us.
And that's the part I find quietly hopeful. We still live in a human world, not a machine one — we want to live well together, and the machines are here to help us get there, not to become the point of it. So the real task is almost the opposite of what most people fear: to step outside all this technology now and then and get strong again at the things that always mattered most, the ones we'd half stopped practising because they'd started to feel less urgent than they are. Knowledge was just scarce enough, for long enough, that we mistook it for the point. The machines are handing that mistake back to us — and what's left when they do is what was always ours.
