This week I was stuck in Frankfurt Hauptbahnhof, sitting in the DB lounge watching trains disappear from the departure board one by one.
Below me, tourists were running around with that particular expression — the one people wear before Germany has fully taught them how Deutsche Bahn works. And I caught myself thinking: I am no longer shocked. I have learned to build buffer time like a survival skill.
That is a problem. Not just with trains.
The Connecting Thread I Did Not Expect
Look at my week on paper and it looks scattered. A university AI class in Lörrach. A coaching session with a YPO master facilitator. A delayed train in Frankfurt. A yes to serving as Network Ambassador at the YPO Global Business Summit in Istanbul.
But when I look closer, every single one of those moments was circling the same question:
What does it actually mean to lead when the system you trusted stops working?
The Deutsche Bahn delay made it literal. But the dynamic shows up everywhere — in AI-first teams where hallucinated citations get treated as facts, in boardrooms where confident output replaces honest conversation, in any operation where we have quietly normalized dysfunction because it has become familiar.
The real leadership skill is not knowing how to perform when things run smoothly.
It is knowing what to do — and who to be — when they do not.
When Systems Fail, Rooms Come Alive
Something interesting happened in that Frankfurt lounge.
The trains stopped. The screens went vague. The announcements meant nothing. And then, slowly, strangers started talking to each other. Someone made a joke about German precision. A tourist asked for directions. Stories about missed connections and accidental overnight stays in random cities started circling the room.
For about twenty minutes, nobody was performing. Nobody was generating output. People were just present.
I do not want to romanticize broken infrastructure — Germany has a serious reliability problem that carries real costs for business and daily life. But that moment reminded me of something I see in high-stakes business rooms too: when the polished system breaks down, you find out whether the humans underneath it have anything real.
For leaders building operational resilience in SMEs and Mittelstand businesses, the question is not just whether your systems work. It is whether your people know what to do when they do not.
Teaching AI to Distrust Itself
Earlier this week I ran an AI Catalyst class at DHBW Lörrach. Local AI setups are still messy — authentication breaks, tools push you toward API costs, and you are reminded that we are still driving the 1913 version of this car: powerful, noisy, and sometimes more mechanic than driver.
But the lesson we built toward had nothing to do with prompts.
We designed a research workflow where multiple AI models investigate the same question independently, then surface contradictions instead of burying them inside one fluent answer. Sources get checked, ranked by citability, and stored with a clean bibliography. The point was simple:
We taught AI to distrust itself.
The most dangerous failure mode in AI-assisted work is not a wrong answer. It is a confident sentence with no verifiable source behind it. In an academic thesis, a hallucinated citation is not a small technical glitch — it becomes an integrity problem. In a due diligence report, an unverified claim is not a minor error — it is a transaction risk.
The students who get this are not the ones who learn the best prompts. They are the ones who understand that AI fluency is really workflow architecture: knowing how to separate research from validation, separate writing from evidence, and never confuse speed with accuracy.
That same discipline applies to every SME owner who is starting to use AI to generate strategy summaries, market reports, or financial narratives. The question is not how fast you can produce the output. It is whether the output is trustworthy enough to make a real decision with.
The Room a Prompt Cannot Replace
The week’s most humbling moment came not from technology but from watching a master YPO coach work.
I am in the process of training as a YPO Game Plan Coach — a facilitation role that helps leadership chapters look honestly at where they stand and work together on where they want to go. It sounds simple. It is not.
A room full of experienced, successful leaders does not need more opinions or more frameworks. It needs someone who can hold the space, ask the right question at the right moment, and help the group hear itself more clearly.
Watching Alan Camilleri work, I realized I was in the presence of a craft I had not yet learned.
In my AI-first world, I live at speed. Agents, workflows, operating stacks, context layers — I am usually moving before most people have opened the tool. But in that room, I was a beginner again. Watching. Taking notes. Quietly trying to understand what made the room feel safe enough for honest conversation.
What I observed was not a method. It was a posture. The great facilitator does not take the room hostage with expertise. He gives his presence and structure so the group can do its own work. If it goes well, the credit belongs to the room.
AI can manufacture strategy texts, polished summaries, and structured recommendations at speed. What it cannot do is hold a difficult silence. It cannot create the moment when a room of experienced leaders decides to be honest with each other. That is human craft. And it requires something AI genuinely cannot replicate: the willingness to be a beginner, to serve instead of perform, and to value the quality of a conversation over the speed of the output.
A Mini-Framework: The Three Rooms
Looking across this week, I keep coming back to three types of rooms that every leader in an SME, scale-up, or transaction process will eventually sit in.
Room 1: The Machine Room AI workflows, data systems, operational dashboards. Here, speed matters. The right discipline is structured skepticism — verify sources, cross-check outputs, build for auditability. Treat AI like cash: use it deliberately, not carelessly.
Room 2: The Strategy Room Owners, advisors, investors, successors. Here, trust matters more than speed. The right discipline is clean governance — verifiable numbers, documented decisions, transferable knowledge. A business that is fit for transaction is a business where the strategy room has earned its credibility.
Room 3: The Human Room Coaching conversations, team moments, the Frankfurt lounge when the trains stop. Here, presence matters most. No output, no agenda — just the quality of your attention. This is where leadership is either real or it isn’t.
The leaders I respect most move fluidly across all three. They use AI seriously in Room 1. They maintain discipline and integrity in Room 2. And they never mistake performance in either of those rooms for the kind of trust that only gets built in Room 3.
What You Can Do Next Week
-
Audit one AI output you currently trust. Pick a summary, report, or recommendation your team generated with AI this month. Trace two or three of its core claims back to their original sources. Note what you find.
-
Name one thing your team has normalized that should not be normal. The Deutsche Bahn effect is real in every organization. Find the dysfunction that everyone has stopped mentioning because it has become routine.
-
Sit in a room as a learner. Identify one person in your network who has mastered something you have not yet learned. Request an hour with them — not to extract information, but to observe how they work.
-
Build a contradiction into your next research process. Ask two people — or two AI tools — to investigate the same question independently, then compare where they disagree. The gap is usually more valuable than the consensus.
-
Before your next important meeting, ask yourself which room you are walking into. Machine Room, Strategy Room, or Human Room. Then calibrate accordingly — different rooms demand different versions of you.
The week that looked scattered on the surface turned out to be one consistent argument: the leaders who will matter most in an AI-first world are not the ones who generate the most output. They are the ones who know when to stop generating, hold the room, and earn the trust that no system can manufacture for them.
That is a harder skill than any prompt.
And it is worth building.
Which of the three rooms — Machine, Strategy, or Human — feels most underdeveloped in your organization right now, and what would it take to strengthen it?