System question instead of tool debate: What the Davos discussion about AI+ and work shows – and what remains unanswered

An ambivalent event

Davos 2026 is over. The limousines have left the valley, the snow is covering the promenade again and everyday life is returning to the boardrooms and ministries. What remains is a vague sense of urgency, coupled with a strange lack of direction.

To counter this lack of orientation, we make a conscious decision in this analysis: We ignore the political noise. The geopolitical squabbles, the rhetorical trench warfare between heads of state on the big stage – we leave all that out. Because the real change took place away from the political show effects.

Anyone who followed the discussions in the economic and technological core of the forum could observe a shift in the tectonic plates. It was no longer – as it was three years ago – about marveling at a chatbot that writes poetry. It was about infrastructure. It was about power. It was about what Satya Nadella called a “platform shift“.

But if you listen carefully, between the lines of the panels on “Jobless Growth” and “Intelligent Co-Workers“, a gap is revealed. A gap that hardly anyone talks about because it hurts. Davos has named the ingredients of the new world of work – AI agents, reskilling, energy transition. But the forum failed to give us a recipe for how leadership should work in a world that can no longer be planned in a linear fashion.

We are at a point where we can no longer see “AI+” as a technology issue. It is a system issue. And it is time to make this classification.

From gimmick to system issue: AI+ is no longer a tool

For a long time, we have clung to the metaphor of the “tool”. AI as the better calculator, as the “co-pilot”. But Davos 2026 has buried this metaphor. When managers like Julie Sweet from Accenture emphasize that it is no longer about “human in the loop”, but about “human in the lead“, this is an admission: technology is no longer just a tool, it is an actor.

We are moving from chatbots to “agentic AI” – systems that act, decide and execute autonomously. This is fundamentally changing the architecture of work. It’s no longer about an employee using software. It’s about the employee becoming part of a hybrid system in which the boundaries between human and machine work become blurred.

Satya Nadella compared this change to the advent of the internet or the PC. But the comparison falls short in one crucial respect: the PC did not act. It waited. The new systems do not wait. They suggest, they optimize, they execute. And this forces us to no longer understand leadership as the distribution of tasks, but as the design of systems. Anyone who only sees AI+ as an IT issue has not heard the message. It is a question of organizational architecture.

The area of tension: productivity vs. meaning

Productivity was a dominant narrative in Davos. The figures are tempting: 14 to 35 percent increase in efficiency in call centers, massive acceleration in coding. But these metrics mask a deeper, philosophical crisis, which Yuval Noah Harari formulated succinctly: If we hand over competence to algorithms, what happens to our identity, which we have defined for centuries by our ability to think?

We run the risk of creating a hyper-productivity that cores meaning. If AI commoditizes the “answer”, what is the question worth? Adam Grant pointed out that while AI is better at generating ideas (volume and variance), humans are (even) better at evaluating and selecting these ideas.

The challenge for managers will be not to fall into the frenzy of efficiency. Yes, we can speed up processes. But if we cut the link between activity and outcome, we risk an alienated workforce that only acts as a “human in the loop” to correct the machine’s mistakes. Productivity increases, but the organizational sense erodes. This is not a technical problem, it’s a cultural one.

Automation vs. responsibility: who is liable for the agent?

Another theme that ran through the halls of Davos was the relationship between automation and responsibility. Alex Karp from Palantir put it in a nutshell: on the battlefield – and increasingly in the market – it is not the PowerPoint presentation that counts, but whether the system works under the most adverse conditions.

We are heading towards a world in which we delegate decisions to “agents” – be it in purchasing, scheduling or medical triage. But responsibility cannot be automated. If an AI agent makes a hallucinated decision, who is responsible? The developer? The user? Or the manager who implemented the system?

There is a massive discrepancy here. We are scaling the ability to act (through AI), but we do not yet have the mechanisms to scale responsibility to the same extent. In the future, leadership will mean taking responsibility for decisions that you have not even made in detail yourself. This requires a radical rethink in risk management and in the ethics of corporate management.

Speed vs. judgment: the end of slowness

The “DeepSeek moment” was often quoted in Davos. It symbolizes the speed at which competitive advantages can shift. What was a moat yesterday (e.g. proprietary models) is a commodity today.

In this world of extreme speed, judgment is becoming the scarcest resource. When information and analyses are immediately available, the bottleneck is no longer knowledge, but the wisdom to do the right thing. We see a polarization: on the one hand an acceleration of processes, on the other an almost desperate search for orientation.

Managers must learn to operate at two speeds: The organization must run technologically so as not to be left behind (keyword: infrastructure development, “race for compute”), but pause humanly so as not to lose direction. Enduring this ambivalence is the new core competence.

The blank space: flying blind as a guide

And here we come to the core of the problem, to what was often only mentioned in passing in Davos. We know what is happening (transformation of work, disruption). We know why it is happening (technology, demographics). But we hardly know how to steer our organizations through this fog.

There is a lot of talk about “reskilling” as if it were a purely technical process – like installing a software update on employees. But the reality, as described by Denis Machuel from the Adecco Group, for example, is more brutal: companies lay off thousands of employees because their skills no longer fit and hire thousands of new ones. This is not “reskilling”, this is replacement.

The blank space in the current debate is the question of transition. How do you lead an organization when the ground is shifting under your feet? How do you motivate people to cooperate with an intelligence that could potentially replace them? How do you plan strategies when 18-month-old predictions are already a waste of time?

Davos has shown: The old maps no longer work. We are navigating a terrain for which there are no maps yet. We no longer talk about AI+ as a feature, but as the operating system of our new reality.

The signal in the noise

Let us end this review with the appeal formulated by Christine Lagarde in Davos, which is perhaps the most important compass for the coming year: we must learn to separate the “noise” from the “signals”. The noise is the hysteria, the daily water level reports and geopolitical provocations. The signal, however, is the fundamental, structural change in work that we need to shape.

For us Europeans, a clear stance can be derived from this: It is time to roll up our sleeves and build. But it is also time to keep our heads up and not allow ourselves to be insulted or talked down to. Those who understand the signal have no time for inferiority complexes – they have work to do.

Our setting

It is no longer enough to marvel at the phenomena or warn against them. We must begin to dissect the mechanisms of this new reality and develop a vocabulary for leadership in this era of “agentic AI” and systemic uncertainty.

In the coming weeks, we will not be looking at the technical specifications of new models. We will be looking at what this shift means for the structure of work, for the responsibility of leadership and for the architecture of our organizations. We are moving away from the discussion of tools to the discussion of systems.

This article is just the beginning. We will not only describe the mechanisms of this new reality, but also dissect them – and underpin them with possible actions. In the coming days, we will publish further analyses and concrete consequences – readable for our registered newsletter readers.