The key element for every AI project: people!

AI promises enormous productivity gains, but has yet to deliver them in many companies. The reason rarely lies in the technology itself, but rather in its interaction with people: their expectations, skills, and daily work reality.

Instead of asking whether AI replaces or relieves employees, the question must be: How does AI empower people in their roles? A human-centered approach does not start with tools, but with tasks, processes, and needs. This is precisely what enables the transition from pilot projects to real value creation – a step that, according to MIT, around 95% of companies fail at.

How can this change be achieved without overwhelming people and with growing trust? That is exactly what this article is about.

Many organisations still view AI as a rival to their employees – a misconception that causes uncertainty and slows down pilot projects. The data shows that projects fail because AI is introduced in isolation from real work processes.

Examples such as CNET, Sports Illustrated and Air Canada illustrate that it is not AI that is the problem, but rather a lack of framework conditions and human control.

A human-centered approach sees AI as an amplifier of human capabilities, not a replacement. Only when expertise, context, and AI work together can real impact be achieved. The supposed dichotomy of “AI or people” thus dissolves: the future lies in cooperation.

Many GenAI pilots fail not because of technical issues, but because of organizational ones: a lack of skills, unclear use cases, and a failure to transfer knowledge into everyday work. Without a shared knowledge base, uncertainty and poor results are the norm.

Pilots often remain abstract, results seem unreliable, and the benefits are not apparent for roles in day-to-day business. In addition, people rarely change the way they work without guidance—training alone is not enough.

Fears, data protection issues, and compliance requirements also inhibit adoption.

The core issue is clear: technology works – but the interaction between people, processes, and AI often does not. This is exactly where a human-centered AI strategy comes in.

Many AI initiatives fail not because they are too ambitious, but because they start with the wrong basic assumption: that technology is the focus. But AI only unleashes its value when people understand it, trust it, and consciously integrate it into their everyday work.

A human-centered AI strategy reverses this perspective. It does not consider the tools as the starting point, but rather the roles, activities, and experiences of the people who work with AI. It asks the question: “How can we support employees in doing their daily work more easily, quickly, and with higher quality?” Which tasks can be relieved? Which new skills are emerging?

This approach differs significantly from traditional “tech-first” programs, which are dominated by rollouts, licenses, and features. A human-centered strategy focuses on something else: the benefits in the real workflow. AI is not simply provided, but designed together with the people who will later use it.

This changes the nature of the entire transformation. Instead of resistance, there is co-creation. Instead of uncertainty, there is trust. And instead of isolated pilot projects, there is a scalable framework that becomes embedded in the organization.

What does that mean in concrete terms? 

A human-centered AI strategy...

• It starts with real tasks, not abstract possibilities.
• It connects technology with context instead of introducing it in isolation.
• It creates clarity and security through understandable rules and transparent guidelines for responsible use.
• It promotes skills and not just tools, so that employees remain capable of acting and confident.
• It makes people active participants in the design process rather than those affected by a change program.

At this point, you may be asking yourself: How can such an approach be implemented in practice? The answer lies in four pillars that enable the transition from pilot project to real value creation, while consistently placing people at the center.

A human-centered AI strategy provides guidance. It offers a clear, repeatable framework for systematically integrating AI into everyday work—with the goal of benefiting people.

Each of these pillars addresses a key hurdle that otherwise slows down AI projects.

1.  Build skills in a targeted manner – based on roles and levels

Many organizations rely on broad training programs that treat everyone the same. However, AI skills do not develop according to the scattergun approach. Different roles require specific skills: from basic AI literacy to risk and validation skills to applied prompting for the respective activity.

It is just as important to meet employees where they are individually – whether they are AI beginners, advanced users, or experts. An effective learning model therefore takes into account not only the role, but also the skill level. This way, no one is overwhelmed, no one is underchallenged, and everyone can build confident and effective AI skills at their own pace.

A role- and level-based learning model teaches exactly the skills that are relevant for a specific position and level of maturity. This allows employees to experience how AI specifically supports their daily work. This lowers barriers, creates confidence, and awakens the desire to continue learning.

You may be asking yourself, “Is that enough for sustainable adoption?” Not yet. Because knowledge alone rarely changes habits.

2. Bringing AI into the workflow – context makes adoption possible

Most pilots fail because they don't arrive in a real work context. A use case often seems convincing in a meeting, but remains invisible in the turbulence of day-to-day business.

A human-centered AI strategy therefore shifts the focus to where work actually happens. It brings use cases, agents, and examples directly to the point of application: into the ERP system, into Office tools, into communication platforms, or into process portals.

When employees see AI exactly when they need it, a natural transition occurs from “I could...” to “I'll just do it now.” In this way, new behaviors are not imposed, but facilitated.

3. Strengthen trust – through simple, visible rules

Trust is the basic prerequisite for any form of AI use. Without clear guidelines, uncertainty arises: What data am I allowed to use? How do I document decisions? When should I deliberately not use AI?

A human-centered AI strategy creates transparent rules that are easy to understand and suitable for everyday use. This includes data protection, governance, and compliance requirements, but also simple “dos and don'ts” – for example, when dealing with sensitive information.

Such guidelines are indispensable, especially in the context of new regulations such as the EU AI Act. They provide guidance, reduce risks, and enable responsible use without fear of mistakes.

4. Make people co-creators – instead of affected parties

The most sustainable changes arise where people take an active role. AI is no exception. When teams can contribute their own ideas, give feedback, experiment, and suggest improvements, genuine ownership emerges.

A human-centered AI strategy therefore relies on local AI champions, communities of practice, and short feedback loops. It transforms those affected into participants—and participants into creators. This greatly increases acceptance and accelerates operational implementation.

Once the strategic principles are in place, the question arises: How does this work in everyday life? The answer lies in the AI lifecycle.

Many organisations treat AI initiatives like traditional IT projects, which are clearly defined and have a defined end point. But AI works differently. It is constantly evolving and changing with every new use case, regulation and update. A static project can hardly do justice to this dynamic.

A human-centered AI strategy therefore relies on a continuous, cyclical approach that is directly aligned with the real workflow. This lifecycle ensures that AI does not end up as a short-term experiment, but becomes a living part of the organization.

Identify: Discover opportunities where work happens.

The first step does not begin in the lab, but in the everyday life of the teams. Which activities take time? Where do bottlenecks arise? Which decisions can be prepared in advance and which information can be obtained more quickly?

Workshops, shadowing sessions, or simple process observations allow meaningful use cases to be identified in a much more targeted manner than top-down lists.

This is where the first added value arises: people feel seen and involved, which strengthens trust.

Prototyping: Test quickly, evaluate reliably

Small prototypes enable rapid insights without great effort. They show whether a use case is practical, what data is needed, and where the risks lie. This phase thus creates clarity before time and budget are invested in extensive developments.

A good prototype answers questions such as:

• Does the scenario work in a real-world context?
• How do people interact with it?
• What validation mechanisms are necessary?

This results in solutions that work in everyday life – and not just on paper.

Scaling: Integrating AI into the workflow, not into a parallel universe

This is where it is determined whether a use case will become a genuine productivity lever. Successful solutions are integrated into the systems that teams already use—ideally with as little friction as possible: directly in the tool, in the process, or in the workflow.

Scaling means not only technical rollout, but above all:

• Contextualization (right role, right environment, right time).
• Support (nudges, coaching, micro-learning) and
• The rollout takes place in waves, tailored to risk and value contribution.

This is exactly where most pilots fail, as they never end up in the “flow of work.”

Communicate: Visibility creates acceptance

AI only gains traction when employees know that new opportunities exist and why they help them. Short, clear communication, practical examples, and specific application tips support its integration.

A simple question helps here: How does a person learn at the right moment that AI can provide meaningful support here?

If the answer to this question is unclear, AI will remain invisible.

Update: Understand usage and think ahead

Real transformation comes from learning during regular operation. Usage data, feedback, frequent errors, or new requirements provide clues as to how a use case can be further improved.

This step creates momentum: AI does not stand still, but matures with every observation.

Improve: From feedback to impact

The final phase closes the loop. Insights are transferred into new versions and the cycle begins again – lightweight, continuous, and reliable.

This creates a pace that doesn't overwhelm teams, but carries them along. The life cycle becomes routine. AI becomes a tool that relieves people in their everyday work, rather than being a separate project that demands attention.

The experiences from countless AI pilots show a clear pattern: technology alone has no effect. It only unleashes its potential when people understand it, trust it, and integrate it naturally into their everyday work. This is exactly where a human-centered AI strategy comes in—creating the framework in which AI moves from being an experiment to providing real relief.

Such a strategy builds skills instead of creating overload. It brings AI to where work happens, rather than creating new portals or parallel worlds. It provides clear, understandable rules that give security. And it makes people co-creators of the transformation, not mere users of ready-made solutions.

When AI is introduced in this spirit, the character of the technology changes. It becomes a tool that reduces repetitive tasks, prepares decisions, and creates time for what humans do best: assessing, designing, collaborating, and developing creative solutions.

The step from using technology to truly benefiting from it always begins with people. When this step is successful, a working world emerges in which AI does not replace but rather strengthens. A working world in which AI does not dominate but rather supports. A working world in which AI creates trust instead of uncertainty.

AI then becomes what it should be: a tool at the service of people and an engine for continuous progress.