"Knowing how to write a good prompt is not enough for workers." That is what Philipp Navratil, CEO of Nestlé since 1 September 2025, told Business Insider in an interview. Navratil leads more than 270,000 employees worldwide, and he does not expect them to master the technical side of AI. What he does expect: that they understand how AI applies to their own part of the business, where the technology is heading, and what that means concretely for Nestlé. It is a message Rescope recognises immediately: ownership over tool knowledge is precisely what Rescope has argued for years. Yet in his first months as CEO, that same Navratil announced that Nestlé is cutting 16,000 jobs. Skills matter more than prompting, says the CEO. Skills for what, and for whom, he does not say.
What happened
Navratil took the helm at Nestlé after the dismissal of his predecessor, Laurent Freixe. Still in his first months as CEO, in October 2025, Nestlé announced it would cut 16,000 jobs over two years: 12,000 office roles and 4,000 jobs in manufacturing and the supply chain. Automation is explicitly part of that strategy, with a savings target of 3 billion Swiss francs (CHF, Nestlé's home currency) by the end of 2027.
About a year later, on 8 October 2026, The Next Web reported on an interview Navratil gave to Business Insider. In it, he calls on his employees to look beyond prompting. He wants them to rethink their own processes rather than simply layering AI on top of the existing way of working. His reasoning: adding AI to an unchanged process often makes the work more complex, not simpler. Only once redundant steps disappear does time free up for analysis and for contact with stakeholders. The focus lies on how the work is designed, not on which AI tool is used alongside it.
Meanwhile, some 100,000 Nestlé employees now use Microsoft Copilot every month, according to chief information officer Chris Wright.
Why this matters
The core of Navratil's reasoning holds, and at Rescope we recognise it from every transformation journey we guide. Anyone who simply lays AI over an existing process rarely gains time. Workflows need to be redesigned and redundant steps need to disappear. That can only happen when the people who do the work every day are themselves in charge of that redesign. That is precisely the ownership Rescope argues for. Not because it sounds good, but because no one knows the process as well as the person who carries it out.
Except that logic raises something the article itself does not examine: what those 270,000 employees are actually given to carry out that redesign. A CEO who calls skills more important than prompting implies an investment in training, time and guidance. The only hard figure the article offers is that 100,000 employees use Microsoft Copilot every month. That is an adoption figure, not evidence of skills built. How many of those 100,000 ever received structured training to rethink their own workflow, rather than simply operate a tool, remains unanswered. A German survey by TU Darmstadt points to how wide that gap usually runs: barely 15 percent of respondents had ever received any form of AI training. Nothing in Nestlé's story rules out that the same gap exists there.
Then there is the sequence. Navratil announced the cuts first, with automation explicitly part of that strategy. Only a year later does he come out with the message that skills matter more than prompting. That does not in itself prove insincerity, but it is striking: if skills really matter more than tools, that insight should have underpinned the savings decision itself, not arrived a year later. The most obvious reading is that Navratil is admitting something here that the article does not spell out. He himself, or his organisation, was not yet AI-literate enough at the time of that decision to put those skills centre stage then.
Decisions of that weight need to be grounded in knowledge, not just followed by a message about skills. That makes AI-savvy leadership just as crucial as AI skills among employees. The Nestlé story offers three points worth testing your own approach against.
Asking for skills without connecting the departments that must decide together remains a one-way street. There is a difference between asking employees to understand how AI affects their role, and actually bringing those employees together with the managers and the steering committee who must make the call. Say an organisation asks its employees to build that understanding, but the decision on which tasks disappear, which steps get merged, or which process gets rewritten sits solely with a steering committee. Without input from the employees and managers who know the work day to day, that understanding stays isolated with whoever must apply it, disconnected from whoever ultimately makes the call. This is not an argument for sidelining the steering committee, a transformation needs direction. It is an argument for bringing employees, managers and the steering committee into the same conversation before that decision is made, rather than a call that forms in one layer and only reaches the others afterwards. That happens concretely when employees help build the framework: why the change is needed, what actually changes, and how it gets carried out. That framework then needs to actually land in a transformation plan, not only in a presentation about intent. Asking for understanding is not the same as connecting the departments that must decide together.
Tool use and skills building go hand in hand, neither proves the other. Without actually working with AI, no one develops a feel for what the technology can and cannot do: tool use is the way in to understanding, not a distraction from it. But the fact that a hundred thousand employees open Copilot every month does not in itself prove they also learned to approach their work differently. As long as an organisation does not measure how much of that tool use translates into a different way of working, the promise of skills over tools remains a slogan without a number behind it. An adoption figure is not a learning curve, but without adoption no learning curve emerges either.
A visible growth path determines whether this is policy or reframing, and that growth path must be tracked across the whole organisation. Anyone who says skills matter more than tools must also be able to show where those skills take employees: a new role, more responsibility, a clear learning path. That takes more than a one-off plan: the redesigned processes that follow from the transformation demand specific skills, and only whoever monitors the progress and orchestration of the plan across the organisation truly knows whether those skills are being built or not. Without that growth path, visible and tracked, "skills matter more than tools" remains a reassuring line with no follow-through. Only a growth path, tracked over time, makes the claim made beforehand credible.
That last point, the growth path, is exactly where Rescope's 3R Platform comes in, developed from Rescope's own practical experience with clients. Through competency matrices and linked training, the platform makes visible what is expected of someone in a role and where that person stands today, so that "skills matter more than tools" does not remain a claim that only holds up on paper. That matrix does not stop at the shop floor: alongside the AI Champion, there is an equally fully-fledged track for the Transformation Lead, so the growth path stays visible and trackable at management level too, not only on the floor where the change is felt first. The AI Champion plays a key role in this: not as an extra layer on top of the organisation, but as the person who helps determine which skills a team needs and who helps shape colleagues' growth path, rather than having a board decide that from a distance. The same division of labour applies when an organisation actually deploys digital employees today, the AI solutions that take over tasks or parts of a role: Rescope manages the technical side, from setup to maintenance, while the substantive choices, which tasks, which knowledge, which decisions, stay with the organisation itself and its Champion. That way, ownership is not only asked of employees: it is also organised.
The real lesson
Philipp Navratil is right that prompting alone is not enough. But the order in which he says so, a year after the decision on 16,000 jobs, says as much as the statement itself. A CEO who today asks for skills without having first shown where those skills lead, is in essence asking for the same thing as before, only in friendlier language.
That is exactly why AI literacy does not stop at the shop floor. It is just as much a competence that needs to grow among managers themselves, not a trait that comes automatically with a job title. A competency matrix that stops at the boardroom misses precisely the level where the knot between cutting costs and investing in training gets cut.
For any organisation investing in AI skills today, the same test applies: not whether the right words about ownership get used, but whether an employee can point today to where the required skills will take them tomorrow, and who stands behind that. That answer should not stop at a role or a learning path. It should also point to which concrete use cases those skills will actually carry, and who takes responsibility for making sure they get carried out.
Can you show that today, for every role in your organisation that works with AI?