Why your team is overtaking you right now

Essay #001 · 2026 · Reading time 8 min · Field: Sovereignty · Market: all

Maybe you know this moment. In a meeting, an employee says something about AI that is more precise than anything you have thought about the subject yourself. You nod and make a note. Later you ask yourself whether you still actually know what is happening in your company.

This is exactly where arocon comes in.

As a managing director, you know where your knowledge carries. You know the market, the numbers, and your customers. You know which conversations you need to have and which you don’t. You built that certainty over years.

In the last two years, however, a topic has arrived that behaves differently: AI agents. You cannot catch up on it over two weekends, because it keeps changing. And the knowledge about it is distributed very unevenly across your workforce, and not along the hierarchy.

In the companies I know, it looks like this: The marketing assistant has been experimenting with agents for ten months and knows more about them than her head of marketing. The young controller runs models at home and argues more precisely than the CFO. The developer builds an internal assistant on the side that works better after one week than the official solution.

What sets AI agents apart from classical tools

A classical tool (CRM, DMS, ERP) is an instrument. You know what you want, and the tool helps you do it. The decision stays with you.

Agents work differently. They are given a goal, not a work step. They decide for themselves which steps to take. They ask when something is missing and deliver a result that is often already usable. In larger setups, several agents work together: one researches, one structures, one writes, one checks.

The consequence: work that used to take hours or days is done in minutes. Not perfect, but often good enough. The rest is rework, not groundwork.

That sounds like an efficiency story. The more important point lies elsewhere: the knowledge about the work is moving. Your people used to know how their work gets done, and you could ask them. Today, a growing share of that knowledge sits in the system and with the people who set it up. If you don’t understand how the system divides up the work, you no longer see where the critical points are.

Why this affects your authority as managing director

Leadership works because, in sum, the boss knows more than any single person in the company. Not more than each specialist in their field, but more in overview. The head of sales was ahead of you in sales, the design engineer in engineering. You had the bigger picture. That overview is the quiet foundation of your authority.

Your authority rests on two foundations. The formal one: contract, company law, the right to give instructions. And the practical one: “My boss knows what he is talking about.” The formal one only holds as long as the practical one does.

When the practical foundation falls away, there is no loud revolt. Employees listen politely and nod. The real thinking, they then do among themselves. You still sign the decisions — they are thought through elsewhere.

I am currently seeing this state in many of my conversations with Mittelstand companies. Not dramatic, not loud, but distinct. And it does not go away on its own: the gap grows every month, because the learning curve at the bottom of the company is steeper than at the top.

One important thing: this is largely not your fault. You did not stop being good. The subject shifted while you kept working the way you always have. But that is exactly why it will not resolve itself either.

Three signals that tell you it is happening

First: proposals you can no longer judge yourself. A department proposes a customer-service bot, an internal knowledge assistant, or a sales assistant. You get a presentation and nod in the right places. In the end you give the green light or turn it down, but you know: you did not make that decision. You delegated without wanting to delegate.

That is no disgrace. It is a signal.

Second: investments whose assumptions you cannot check. Your head of IT proposes an infrastructure package: vector database, embedding models, orchestration, monitoring. A mid six-figure sum, the numbers cleanly calculated. But you cannot verify the underlying assumption that all of this will still matter in three years. The proposal itself won’t answer that, because it was written by the people who like it.

In the past, you would have asked your advisory board. The boards I encounter, however, are rarely deeper into this subject than you are.

Third: staffing decisions without reliable experience. In a department with intensive agent workflows, a staffing question comes up: does this role disappear, does that one grow, is a new one needed? The manager has one opinion, you have another. Who is right cannot be settled by experience: for this situation, there is none yet.

At this point you may notice for the first time: my opinion is not resting on my own knowledge right now, but on reflex. That, too, is a signal.

Four common answers and why they don’t solve the problem

Various answers are on offer for this situation. In my view, none of them solves the actual problem.

AI training for the team. The team is not the problem; it is part of the solution. Training treats the symptom, not the cause. And it costs you the time you would have needed to get moving yourself.

An external Chief AI Officer. A new role with AI responsibility merely shifts the problem. The CAIO stands between you and the decisions you have to make. Instead of asking your developer, you now ask a consultant. Other people’s knowledge advantage remains, just with an official mandate. That is another delegation, not a solution.

A consultancy’s strategy deck. A good deck is reassuring: it orders, prioritizes, recommends. But it does not replace your own thinking. What you need is not the deck but the ability to check it yourself.

A broad tool rollout. Tools come later. The question is not which copilot package to buy. The question is how work in your company will be divided between people and agents in the future. That is an organizational question, not a product question.

All four answers share the same core: they try to solve the problem with outside expertise. But that is exactly the mechanism that created the problem.

What helps: learning it yourself, a few hours per week

The answer that works is uncomfortable: you have to learn this yourself.

You do not have to become a developer or configure agents by hand. The point is to be able to ask the questions nobody can answer for you:

• How does our value creation change when thirty percent of today’s activities are done by agents?
• Which competencies lose weight, which gain?
• Which roles do we need to replace, reshape, or create?
• What may agents decide, what not, and who checks that?
• Which investments are necessary, which are fashion?
• What does liability mean when decisions are no longer made by humans alone?

These questions are not technical. To answer them, you do not need to be able to code. You need to understand the logic by which agents break down and complete work: the principle, not the configuration.

The effort is more manageable than you might fear. A few hours per week, consistently over three to six months. After that, you can judge your people’s proposals fairly and decide from your own judgment again instead of trusting someone else’s recommendations. Not because you would then be smarter than your employees, but because you can once again see where value is created in your company, and how agents amplify or shift it.

The first step takes two hours: sit down with the employee who is furthest ahead in your company and have them show you a real agent workflow. Not as a presentation, but on screen. Afterwards you will know where you stand. From then on, block one fixed slot per week for the subject.

arocon accompanies this path: confidential 1:1 sparring, monthly essays, small groups of managing directors. No course programs, no certificates, but a working relationship for decision-makers who want to get moving themselves.

Because the lead belongs where the decisions are made: with you.

— Axel Roth