Interventions

What remains human in the generational transition of a company in the age of artificial intelligence

by Alfredo De Massis and Vittoria Magrelli

9 May 2026

3' min read

Translated by AI

Versione italiana

3' min read

Translated by AI

Versione italiana

There is a word that Roman law understood better than we do: heres. We translate it as 'heir', but the Indo-European root from which it comes, ghe, grasp, says something more demanding. The heir is not the one who receives: it is the one who grasps. An active, almost physical gesture, implying will, tension, risk. Inheriting, in its deepest origin, is not a state but an act.

For centuries, family businesses have lived within this logic. The founder built, the heir grasped: relationships, tacit knowledge, reputation, market flair. A transfer that took place by osmosis, by proximity, by slow and conscious imitation. The numbers say that this season has reached its most critical moment. In March 2026, the Young Entrepreneurs of Confindustria estimated that almost half of Italian family businesses will face a generational change within the next decade. This is not a distant transition: a transition already underway, affecting the productive heart of the country.

Yet the public debate almost always misses a crucial question: what does broadcasting really mean? Not in the fiscal or corporate sense, which does matter, but in the deeper sense. What was once grasped when inheriting a business? And what changes when a non-human interlocutor enters the transmission channel for the first time in history?

The Latin transmittere does not simply mean 'to give'. It means to send across: time, generations, a medium. For millennia, that channel was exclusively human: the body, the voice, memory, presence. Today, artificial intelligence introduces a non-human channel into this chain for the first time. It is not a question of replacing the heir with an algorithm, but of something more subtle: the possibility of encoding and making consultable a part of that knowledge which until yesterday was by definition intransmissible.

Think of the founder of a manufacturing SME who has managed relationships with key suppliers for forty years. He knows when to trust and when not to, he recognises the signs of crisis before they emerge in numbers. This knowledge, made up of patterns, exceptions, settled intuitions, has always been the most valuable competitive advantage and at the same time the most fragile: destined to dissolve with it. Today, advanced AI tools can analyse decades of communications and decisions, returning a map of that knowledge. Not the knowledge itself: its trace. Not intuition: its grammar. And they pose a question to the generational transition that has never been asked before: if some of the founder's knowledge can be captured by a machine, what specifically human remains in the succession?

The answer lies in the etymology of a third word. Intelligere, from which 'intelligence' comes, literally means to read between, inter-read, to grasp what is not written. AI is extraordinarily good at reading between the lines, less good at inter-reading. It does not have access to what has never been said: the conversation the founder chose not to have, the client abandoned because something suggested he or she should, the relative turned away because of a present or past family history. It is in this space, the unspoken, the uncodifiable, that lies the most authentic core of what an heir must grasp. Not data: judgement. Not memory: the capacity to form new memory.

The heir entering a family business today thus finds himself in a paradoxical position: he has access to more explicit knowledge than any predecessor in history, while at the same time having to develop a capacity for judgement that no platform can ever replace. The risk is that the abundance of data becomes an alibi for avoiding decision fatigue, that the map is mistaken for the territory.

Hereditas, in the end, is not an archive. It is a gesture. And gestures are not transmitted through a digital channel: they are learnt by being close to those who know how to do them, by observing, by making mistakes, by trying again. AI can hold the trace of those gestures. But grasping them, in the full, ancient, demanding sense of heres, remains an irreducibly human task.