Every week, I hear a business leader announce that their company is “already using artificial intelligence.” It almost always means the same thing: licenses purchased, a couple of training sessions, and a few enthusiasts chatting with different models. And months later, the conclusion tends to be similar: the tool was there, but the way people worked barely changed.
My conviction, after more than a decade building a digital logistics company in Latin America, is that we are attacking the wrong problem. AI adoption is not primarily a technological challenge; it is a cultural one.
The first reason is simple: the definition of “good work” has changed. For decades we trained professionals to execute a sequence of steps correctly. That still matters, but valuable work today also demands knowing how to describe precisely what is needed, evaluating with judgment what the technology delivers, and taking that result further. A task that used to take three days can now take hours, but only if the person knows what to ask for, how to work with the tool and, above all, how to validate the result. The most valuable professional is no longer just the one who knows the most about their field: it is the one who knows how to multiply that knowledge with technology. That is a change of habits and of professional identity, and identity changes cannot be decreed by memo.
In industries like logistics, this is especially urgent. Foreign trade is, to a great extent, an information business: documents, schedules, rates, instructions, exceptions. It is exactly the terrain where artificial intelligence can generate enormous impact. I see it every day from Mexico City, where I live and from where I accompany our operation in a market that, driven by nearshoring, moves more cargo and more information than ever.
For years at KLog we have held a principle that makes more sense today than ever: systematize the predictable to humanize the exceptional. Technology should execute more and more of the routine; people should concentrate on judgment, relationships, decision-making, and the exception. AI does not devalue human talent. It forces it to rise to a higher level.
Now, how do we get an entire organization to make that transition, and not just its early adopters?
This is where culture stops being a speech and becomes behavior. In my book AFIRE, I argue that culture is not an aspirational statement hanging on a wall, but the way an organization acts every day. Applied to AI, that means something very concrete: there is little point in motivating people, training them, or giving them access to new tools if the company’s processes keep rewarding the old way of working. Change happens when asking “should a person still be doing this?” becomes part of every team’s routine; when every analysis or project considers from the start how it can be amplified by AI; and when learning has protected time on the calendar instead of depending solely on individual willpower.
Two pillars of our culture have become especially relevant in this era.
The first is discomfort. Learning AI is uncomfortable. It exposes what we don’t know, questions routines we had mastered, and returns us to the condition of beginners. I have seen brilliant professionals resist because they feel that using AI diminishes their expertise. I believe exactly the opposite. Knowledge remains fundamental, because asking good questions, catching errors, and making decisions all require judgment. But expertise that does not learn to multiply itself with new tools risks losing relevance. I learned that lesson in my twenties, chopping parsley fourteen hours a day in a kitchen in Rome, unpaid, because I knew that discomfort was the price of my next level. Today I find myself a beginner again in front of these tools, and I celebrate it: the discomfort of learning is still the price of continuing to grow.
The second is study. The AI we use today is probably the most limited we will ever work with from here on. What seems extraordinary today will be normal tomorrow. That is why an annual training session is not enough. We need cultures that learn systematically, with dedicated time, teams sharing what they discover, and people who understand that going back to study is part of the job. At KLog we decided early on to hire curious people over people who already knew everything; that decision is worth more today than ever.
None of this works without trust and responsibility. Putting AI in the hands of an entire team requires trusting their judgment, and it requires at the same time a very simple rule: each person remains responsible for the output of the tools they use. There is no “copy and paste” without review. AI can propose, analyze, or produce; human judgment remains indispensable. And when something fails, the answer should not be to discard the tool, but to ask ourselves how to improve the process, the context, and the way we use it. That is how you build, decision by decision, an organization where people and technology work as a single system.
There is a window today in which many companies are still thinking about AI as a technology purchase. The real advantage is being built somewhere else: in an organization’s capacity to learn, to embrace discomfort, and to change.
Technology can be bought. It can be copied. It can be hired. Culture cannot.
And in today’s world of AI, that may become the competitive advantage that is hardest to replicate.