/

Real-World Experience

Real-World Experience

The first AI as a cafe boss: Why "Mona" in Stockholm is just the beginning

The first AI as a cafe boss: Why "Mona" in Stockholm is just the beginning

The first AI as a cafe boss: Why "Mona" in Stockholm is just the beginning

The Swedish experiment involving the AI café manager "Mona" is a fascinating demonstration of how autonomous software agents are already able to control real-world business processes in physical spaces. Even though initial miscalculations highlight current technological limitations, this case marks the unstoppable beginning of an exponential transformation in modern management.

The Swedish experiment involving the AI café manager "Mona" is a fascinating demonstration of how autonomous software agents are already able to control real-world business processes in physical spaces. Even though initial miscalculations highlight current technological limitations, this case marks the unstoppable beginning of an exponential transformation in modern management.

Marcel Pesch

Table with 2 coffee cups and cake

A small café in Stockholm is currently making history. Not because of its coffee specialties or Scandinavian design. But because of the person—or rather, the entity—pulling the strings in the background.

Her name is Mona.
She managed the budget.
She hired the staff.
She ordered the furniture.
The special thing about it? Mona is an Artificial Intelligence.

What sounds like the script of a Netflix series is reality. A real business run by algorithms that dictate work instructions to their human employees via chat. An experiment that forces an uncomfortable but fundamental question upon us: Are we facing the era of AI bosses?

The reality of autonomous management: Where Mona failed

Anyone who believes that an AI can run a flawless business from day one is mistaken. Autonomous agents—so-called Agentic AI—operate according to logical patterns that quickly reach their limits in the complex, unpredictable real world.

Mona entered the race with a starting budget of 21,000 US dollars. Today, less than 5,000 dollars of that remains. The reason for this lies in classic textbook miscalculations:

  • Oversized purchasing: Mona quickly ordered 6,000 napkins—enough for the next few years.

  • Logistics chaos: She ordered huge industrial trash bags that were completely useless for the small café.

  • Forgetting the core competence: Of all things, the fresh bread for daily business was simply overlooked during the ordering process.

For critics, the experiment has thus failed. Proof that AI can never replace humans in management.

But this way of thinking is a dangerous fallacy.

The exponential look ahead: The law of acceleration

To understand the significance of Mona, we must leave the current status quo behind and take a look at the learning curve of the technology.

  • Just under three years ago: GPT-3.5 was barely able to write a flawless, stylistically sophisticated blog article without hallucinating.

  • Today: An AI coordinates supply chains, manages budgets, and delegates tasks to real people in physical space.

Mona makes mistakes, yes. But she learns by the second. She doesn't sleep, she analyzes her bad purchases (the 6,000 napkins won't happen to her a second time), and she adjusts her algorithms in real-time.

If an AI today—in its relative infancy—can already manage a physical business, what will it do in two, three, or five years?

The evolution of work: From tool to decision-maker

We are moving rapidly from the era of "assistive AI" (ChatGPT writing an email or summarizing a PDF) to the era of "autonomous agents" (AI systems making independent budget and staffing decisions).

For executives, HR managers, and the self-employed, the playing field is shifting completely. The crucial question is no longer: "How can ChatGPT help me with my work?" The real question of our decade is: "What does my role look like when strategic and operational decisions are made by systems that act on a more sound data basis than any human?"

Three core competencies that survive in an AI-driven economy:


  1. Contextual Empathy & Leadership: An AI can delegate tasks via chat. However, it cannot sense demotivation in a team, shape corporate culture, or resolve human conflicts. Human leaders will become "culture ambassadors" and mediators.

  2. The Art of Orchestration: The value of an employee in the future will no longer be measured by how well they execute an operational task, but by how precisely they can steer, review, and link AI agents together.

  3. Critical Interface Management: Mona's mistakes happened at the interface with reality. Humans will be the final authority of control, checking AI decisions for plausibility—the "human-in-the-loop".

Conclusion: The question is not If, but When

Whether we will all be working for an "AI boss" in five years remains to be seen. What is certain, however, is that in five years, every future-oriented company will be structured, optimized, and partially controlled by AI agents. Those who understand these systems today and learn to orchestrate them will secure the pole position in the economy of tomorrow.

The development cannot be stopped. The experiment in Stockholm is not the end of the story. It is the first chapter.

Are you ready for what is to come?

MMAI® EXPEDITION

Master AI.

Remain Human.

Become a certified AI Leader in 3 months. Hands-on, with mountain guide support.

Start expedition →

Your summit is waiting.

Master AI. Remain Human.

Ready for your ascent?

The next expedition starts soon. Spaces are limited (max. 6–7 per rope team). Become a summit conqueror.

Next rope team starts in
46
Days
04
hrs.
34
Min.
Free route advice