Zum Inhalt springen
Essay #003 · 2026 · Reading time 7 min · Field: Personnel · Market: Mittelstand/Trades

Not positions are replaced but activities. Three archetypes shrink, two grow, one strategic role emerges anew. The personnel arithmetic of the age of agents is an entrepreneur's task — not HR's.

Whom to hire now, whom to let go

Essay #003 · 2026 · Reading time 7 min · Field: Personnel · Market: Mittelstand/Trades

One of the most uncomfortable moments in the life of an entrepreneur is the conversation in which you tell a long-standing employee that their role is no longer needed. You do not hold it often. You never enjoy it. And when you do hold it, you usually have the feeling that the person in question has done nothing wrong — rather, circumstances have shifted.

In the coming three years, many entrepreneurs will have to hold this conversation more often. Not because agents “destroy jobs” — that is the wrong framing. But because agents shift the requirements of individual roles to a degree where certain constellations no longer fit.

This essay is about the personnel arithmetic you have to run from now on. It is about which roles shrink, which grow, which emerge anew. And it is about why that arithmetic is not an HR task — but an entrepreneur’s task.

The wrong framing

Let us begin with what this essay does not say. It does not say: agents will replace thirty percent of positions in Mittelstand companies. That number circulates but is misleading. Not positions are replaced, but activities. The distinction sounds academic, but it has consequences: a position usually consists of many activities, and if a third of these activities become automatable, the position has not disappeared — it has changed.

The decisive question is not how many positions fall away. The decisive question is: which role cuts still carry in five years? Which do not? And what does that mean for your current hires?

Three shrinking archetypes

The mediator between systems. People whose main activity consists of moving information from one system to another — order-processing clerks, administrative assistants, parts of controlling. These roles were already thinned out by digitalization over the last fifteen years. The remaining work was what the old systems could not do: complicated cases, exceptions, checks. Precisely that work is what agents can now often handle. What remains is less than the people in it themselves believe today.

The long-text producer. Marketing copywriters, offer writers, report authors — roles whose core service was turning bullet points into tidy prose. Prose, agents can now do themselves. What remains is the curatorial work: deciding which text is really needed, which tone fits, which position is taken. That is less hand, more head. Most people in these roles have been doing the head-work already — but not primarily. Primary was the hand-work. And that gets thin.

The competence-island holder. The most dangerous category because least obvious. Employees whose value lay in the fact that they had specialist knowledge no one else had — a complex set of rules, an experience-based procedure, a historically grown customer relationship. Agents break open knowledge islands. What a person accumulated over years can become available in days in a well-fed system. That devalues the island. Not overnight — but distinctly.

Dealing with the third archetype is especially delicate because the people in it often sit in key positions, have been loyal for a long time, and find it hard to surrender their knowledge to the system voluntarily. That is understandable. It is also a risk to the company if those islands are not broken open.

Two growing archetypes

The judgment worker. People whose value consists of making good decisions in complex, uncertain situations — senior engineers, experienced product managers, experienced customer advisors in specialist segments. Agents feed them, but they do not replace them. On the contrary: the value of a judgment capability rises when more material can be processed beneath it. A senior with agentic support delivers what used to require a group. That makes these senior roles more valuable, not less.

The orchestrator. People who design, monitor, and correct agentic systems. This is not a purely technical role — it is a hybrid that combines domain-level process understanding with technical agent intelligence. In every company with more than fifty employees, three to five such roles will emerge in the coming years. They will be well paid, and they will be scarce in the market.

A newly emerging role

In addition to these shifts, a role emerges that did not exist before: the person whose task it is to lead the company’s dealings with agents strategically. Not a Chief AI Officer in the classical sense — and not the same CAIO the first essay of this archive warned about. There it was about an external advisor who stepped between you and your people. Here it is something else: a role that emerges internally and holds the connection between business strategy and agentic infrastructure.

In smaller companies this role is often part of leadership itself. Not delegated but integrated. In mid-sized companies, a feeder role emerges directly alongside the executive team — not at IT level, but at strategic level.

What you call this role is secondary. That you fill it is primary.

The arithmetic you must now run

When you make a hire today, ask yourself a single question: Does this role still belong to the company in five years? If the answer is “yes,” hire. If the answer is “I don’t know,” think twice. If the answer is “probably not,” do not hire — but invest the budget into one of the growing archetypes or into the newly emerging role.

That is not an absolute. It is a compass. Not every single hire has to be future-perfect. Some roles you need today even though you know they will look different in five years.

But the compass helps you stop making personnel policy out of habit, and start making it out of strategy.

The hard conversation

And what about the roles that shrink? What about the employees who have been here a long time, who are good, who have done nothing wrong?

There is no comfortable answer. Part of these roles can be reshaped — the clerk becomes a reviewer for difficult cases, the copywriter becomes a curator. That requires training, and it requires that the person is willing and able to go along with the change.

Another part cannot be reshaped. Then it is about designing the transition with decency. Speak early. Compensate fairly. Give references. Do not wait until the pressure is great enough that you must be curt.

The unpleasant point: those who postpone these conversations too long make them more painful in the end, not milder. The cost of waiting is not borne by you — it is borne by the employees who have fewer transition options with each passing month.

The actual point

The personnel arithmetic in the age of agents is not an HR task. It is an entrepreneur’s task. Because it requires someone in the company to know the strategic direction — and you know the strategic direction if you took the first essay of this archive seriously.

If you do not know the direction, you cannot run the personnel arithmetic. You will only treat symptoms. A new hire here, a severance there. And at the end, you will find that you have moved a lot of money and nothing has changed structurally.

Those who know the direction do the arithmetic once, correctly. And then have the difficult conversations with a clarity that is better for the employees than any avoidance.

— Axel Roth