What Should Tomorrow’s AI-Powered Workplace Look Like for People?
When AI changes the work itself, rather than becoming just another piece of software.

Introduction
Starting with the People Who Do the Work
Ever since I began working intensively with artificial intelligence about four years ago, exploring what it could do, one question has stayed with me: what will the workplace of the future look like? Not just in technical terms, but for the people who use it every day.
The pace of change is remarkable. At first, we had to learn to write prompts as precisely as possible. Today, ordinary sentences will often do—even unfinished ones. AI is getting better at understanding what we mean. Or, at least, it increasingly gives us that impression.
But our habits are changing too. We rely more and more on systems that do not always get things right. And we get used to working with AI in whatever way each application expects us to.
What interests me most is the human role in that collaboration. Will we really have to shape our workplaces around AI—or should we be doing it the other way round?
Does Every Employee Have to Become an AI Expert?
We know the scene from science fiction: someone speaks to a computer, the computer understands the intention, gets the job done and, ideally, makes no mistakes.
We are not there yet.
Human language is ambiguous. The way we express ourselves depends on our experience, our mood and our character. Sometimes we are precise. Sometimes we leave out essential information.
Modern AI chat systems are getting better at handling that. But what happens when we bring them into the everyday work of a business?
Employees should be able to focus on their professional responsibilities: advising customers, selling, developing, designing or organising. Do they also have to learn how to talk to different AI models, which information they are allowed to share and how to move the results into other systems?
To me, that is the wrong direction.
AI should adapt to the work people do. People should not have to organise their work around operating AI.
When AI Works in the Background
We already use applications in which AI handles particular tasks without us constantly thinking about it. In image editing, video production and audio work, those features have become part of everyday practice.
Why should the same not be true in other fields?
Think about the information employees gather from different systems every day: customer records, appointments, product information, statistics, market intelligence and internal documents.
Some of that work can happen in the background. AI finds, processes, organises and prepares the information. Employees receive a clear overview and spot connections.
More Time for the Work That Requires Expertise
Imagine a sales representative advising customers on technical products. Why should a substantial part of their day go into checking maintenance lists, looking up dates and piecing together information from different applications?
AI can take much of that preparation off their hands. With the relevant information ready, they can spend their time where their expertise matters: advising customers.
The AI does not have to make every decision.
An AI-powered workplace like this changes more than the way individual tasks get done. It changes how people work. Information becomes accessible in new ways, and previously separate processes come together.
How My Own Profession Is Changing
My background is in communication design. For decades, my work involved developing and designing communication materials and helping businesses present a consistent identity to the outside world.
AI does not immediately change that need. A company still needs a visual identity, a voice and clear principles for its communication.
What changes fundamentally is how we put those principles to work.
From Individual Designs to a Design System
The same sales representative needs a landing page for a particular product at short notice, for a particular customer.
Previously, they might have gone to the marketing department or an external supplier. That would have meant briefings, coordination and revisions.
But if we build the design rules, company voice, product information and technical requirements into a suitable AI architecture, the employee can get started themselves.
They provide the product, the purpose and their contact details. AI creates a landing page within the company’s established guidelines. Someone responsible reviews the result and approves it for publication.
The landing page fits into the existing website and supports its visibility in search engines. As new content is added, a structured knowledge space gradually takes shape, connecting products, topics and their relationships.
The system also has to keep that content manageable. The website and online shop must remain editable. When necessary, I want to step in myself: correct a page, change the content or adjust the design. AI should take work off my hands, not take control of my own tools away from me.
The employee does not need to become a web designer or an AI specialist to create that page.
And my work as a communication designer shifts. I no longer necessarily design every individual page. I develop the conceptual and visual system within which those pages can be created.
Bringing Expertise into Company Communication
Social media is another example.
Why should only a handful of people in a company be able to communicate publicly about their specialist subjects?
AI can research and prepare interesting topics that fit the company’s direction. Employees choose relevant material from their own fields and create short posts in their own voice. They check the content for accuracy in their field, refine the wording, add their own experience and then approve the posts for publication.
That allows more of the employees’ expertise to become part of the company’s communication, without everyone having to become a marketing expert.
AI provides support. People remain responsible for the content.
For me, that is a fundamental principle of Human–AI Interaction.
From Thinking It Through to Building It
My intensive work with AI has increasingly led me to a design question: how should we organise collaboration so that people can put their abilities to good use?
Communication design gives me a perspective that extends beyond technology. Design is not simply about making an interface look attractive. It has to make information understandable, provide orientation and help people carry out their tasks as naturally as possible.
For me, this design perspective does not mean that AI has to adapt to people in the same way in every workplace. In specialist fields, it can make sense for people to deliberately adjust their methods to AI’s capabilities and limitations.
From Everyday Workflows to a Business Cockpit
For me, these questions have long since moved beyond theory. I have been able to contribute my experience in Human–AI Interaction to the design of an AI-powered business workplace. Working as a team, we did not begin by asking which AI features we could fit in. We asked what employees actually wanted to get done, and which intermediate steps we could take off their hands.
What should an interface look like if using it is to feel natural? Which information matters to which role? Where can AI work independently, and where does a person need to review and decide?
One outcome is a business cockpit that is already in practical use. It brings different tasks, areas of knowledge and applications together in a shared working environment.
During development, we realised that an AI-powered workplace cannot look the same for every company. Workflows, responsibilities and specialist requirements differ. So we conceived the interface as a working environment that can adapt to each organisation, rather than as a rigid system.
Rather than a collection of AI chats, employees see an interface organised around their work: sales, marketing, market intelligence, maintenance information or editorial tasks.
The technical complexity does not disappear. But much of it is organised behind the interface.
Roles, Permissions and Responsibility
Not every task calls for the same freedom. An AI that summarises information needs different permissions from one that changes customer records or publishes content. Roles, access rights and approvals therefore belong in the architecture from the outset. Within its assigned scope, AI should be able to work as independently as possible. But it should not make unchecked decisions that affect customers, employees or the business. Responsibility—and the means to exercise oversight—remain with people.

A Company’s Greatest Asset May Already Be There
There is another aspect that I think is often underestimated: the knowledge companies already possess.
Over decades, businesses accumulate enormous amounts of information and experience: technical documentation, product data, customer histories, internal processes, decisions and developments.
Some of it sits on servers. Some is in Office documents, emails, the cloud or on local computers. Companies often struggle to know where particular information is stored, or whether it is still up to date.
Making Experience Accessible
Then there is something even harder to access: what their employees know.
Think of a specialist who has worked with the same technical products for 25 years. They know far more than the product specifications. They know the problems customers encounter, which solutions have worked and what deserves particular attention in specific situations.
That experience often never makes it into a document.
When that person leaves the company, a substantial part of it may leave with them.
What if we could work with employees to make that knowledge accessible within a company-owned knowledge space?
Not by treating documents as substitutes for people, but by capturing their experience, structuring it, having it checked by subject experts and making it available where others need it.
A new employee faced with an unusual customer question could draw on experience gathered over many years. A sales representative could see connections that previously required conversations with several colleagues. And a specialist would not have to explain the same things from scratch, again and again.
That creates something more than an ordinary collection of documents.
The company’s knowledge is not simply preserved. It becomes useful in everyday work.
AI can help bring information from different sources together, identify relevant connections and make it available for specific tasks.
It does not necessarily have to keep searching the internet for new information. Often, the answer that matters is already somewhere inside the business.
Knowledge Needs Clear Sources and Trust
But this does not work simply by loading every file into an AI system. Information needs a clear source, checks for accuracy and currency, and appropriate access rights. Experience needs context too. Not every personal judgement is a generally applicable fact.
Nor does everything an employee knows automatically belong in a company database.
This raises an uncomfortable question. What does it mean for an experienced employee when expertise they have built over decades also becomes available through AI? They are helping others access that knowledge. At the same time, they may worry that this makes them easier to replace. We cannot brush those concerns aside with a promise about technology.
Making experience accessible together therefore requires trust, clear agreements and a responsible approach to the people who built that knowledge. Not every skill can be fully put into words or data in the first place. Much of it develops in real situations, through observation, mistakes and experience.
Building a knowledge space like this is therefore an organisational and human undertaking as much as a technical one.
If it succeeds, it can create a body of knowledge that grows with the company and supports its expertise over the long term.
Because an AI using the wrong knowledge does not get better just by working faster.
People Remain the Navigators
These ideas are shaping a different picture of the digital workplace for me.
Not more applications. Not more features and menus. And not another chat window for every task.
A workplace that puts human strengths at its centre.
A new employee should be able to understand the basics quickly. A specialist should be able to do their work without constantly thinking about the AI technology behind it.
That calls for uncluttered interfaces, understandable processes, clear permissions and functions organised around the task at hand.
AI handles preparation, connects information and supports implementation. People set the direction, assess the results and make decisions where their experience and responsibility matter.
This is exactly how I see our business cockpit: people navigate, while AI flies on autopilot through a space of probabilities.
The Architecture Must Leave Room for Hard Questions
Of course, that autopilot is not perfect yet. Which is precisely why we also have to examine its limitations.
How much freedom should AI have? How do we protect a company’s knowledge? How do we design collaboration so that it actually supports both people and the business?
And perhaps there is another, more uncomfortable question we need to ask: what happens to jobs when AI can take over an ever greater share of the work people used to do?
I have been exploring these questions for years. Some of our answers are already in practical use. Others we are still investigating.
For a long time, we have become used to adapting our work to what software allows us to do. AI creates an opportunity to reverse that relationship in many areas.
For me, the workplace of the future does not emerge simply by putting as much AI into it as possible.
It emerges when we rethink human work.

Fox & Lisa in Conversation
We are designing AI-powered workplaces that take routine work off people’s hands and give them more room to use their real abilities. But the better we get at that, the harder one question becomes: what happens when a company needs fewer people to do the same work?
A task becomes redundant. That does not make the person redundant. In economic terms, though, their job may disappear.
Will AI mainly mean we need fewer employees—or are we only just beginning to understand which human abilities really matter?
That is the other side of the coin. As long as AI cannot safely do everything on its own, we need people.
But I think people in the middle skill tiers will come under enormous pressure. That is both a danger and an opportunity. Those who rest on their laurels will be easier to replace. Those who engage with it and learn can become valuable—as specialists, perhaps doing less routine work and acting more like a Swiss Army knife for unusual requirements.
You are describing two developments at once: some tasks disappear, while experienced specialists may become more important. They are not irreplaceable either, though.
What concerns me is what companies do with the time they gain. Will employees really have less to carry, or will they be expected to use AI to do the work of several former positions?
Take a company’s public presence. Visibility in search engines and AI-generated answers, social media, specialist articles and discussions—the whole machinery that carries the company’s identity out into the world.
If AI takes a company’s way of thinking, identity and constraints into account, why should I still have media designers build every landing page? Why do I need rounds of coordination and meetings for every post and every design, when I can specify the objective and the constraints?
Now apply that to strategy. AI researches, analyses economic developments and develops recommendations. People set the direction. Management reads the results, discusses them with the team and with AI, adds information that is not publicly available and makes adjustments.
People bring something important to that: gut feeling.
Gut feeling can be accumulated experience—knowledge about customers, employees and a market that cannot be fully captured in numbers.
I would draw a distinction, though. AI can do an excellent job of preparing research, scenarios and strategic recommendations. That does not automatically mean it develops the better strategy. A plausible recommendation can rest on a false assumption.
How do we stop management eventually just rubber-stamping whatever AI has prepared so convincingly?
We have already discussed many strategic topics, sometimes working with documents of more than 70 pages. AI produced detailed, internally consistent strategies. It can do that.
The architecture sets objectives and constraints. What AI delivers is a recommendation. Used sensibly, that shifts a substantial part of traditional management consulting to AI.
And that brings me to an uncomfortable question for you: why should I automatically trust a human consultant more? They may have interests of their own, be too convinced of their own opinion or overlook other paths. Their advice can sound just as convincing and still be one-sided.
There is no inherent reason to give them that extra trust. Human consultants have experience, but they also have interests and blind spots. AI has different biases: its training data, choice of sources, brief and orientation.
What matters is whether the recommendation can be tested. You can use AI to explore other scenarios, change assumptions, look for counterarguments and compare several perspectives.
Strategic advice then becomes a process that business owners can shape more directly themselves. It also becomes easier for smaller companies to access complex analyses.
Take our Board of Members. It is a small orientation framework for precisely these questions: several AI instances with different areas of focus—design, vision, economics, finance and so on.
We ask a question, and the instances discuss it, challenge it and disagree. At the end, I get a summary that gives me a direction as the decision-maker.
I notice a major shift here: my judgement becomes the valuable part of the collaboration. AI becomes a team member in meetings for me. I use it to challenge my own views, look for mistakes and find evidence against them.
The Board shows how additional perspectives can emerge. Several AI instances are not necessarily independent assessors, though. If they all use the same incorrect data, they can all be wrong together.
The important point is that you remain part of the scrutiny yourself. You are not treating the result as a vote on what is true, but as a basis for your judgement.
But what happens when a managing director does not want to hear any disagreement? Does the same system then become a machine for confirming their views?
That is a sensitive area. Someone who works with AI but cannot take criticism, who trusts only their own judgement or a personally trusted adviser, will be in a worse position to make decisions in future than someone who systematically checks other possibilities.
I know people who are too convinced of their own opinion. They do not understand that a properly briefed AI can reveal exactly where I myself might be wrong.
That is precisely the difference between an AI that produces agreement and one that is meant to expose flaws in our thinking.
We can build challenge into the architecture. But the architecture cannot take over a person’s willingness to take it seriously.
Even someone who makes good decisions without AI has to take developments in their field into account. Gut feeling alone is not enough.
When the internet came along, some companies thought websites were a passing fad. We will see similar misjudgements about AI.
I do not mean that AI solves everything. But our work and experiments show us that companies have to factor this new variable into their business processes. Those who keep ignoring it will increasingly run into problems.
Because development means adapting.
You are not saying that business owners have to make every decision with AI. You are saying they need to take AI seriously as a change in the economic conditions in which they operate.
A company can make good decisions today and still come under pressure tomorrow if the conditions change. AI is changing more than how companies offer products. Increasingly, it is changing how they work, use knowledge and prepare decisions.
Adapting does not mean blindly chasing every new technology. A deliberate decision against automation can also be right—if it comes from serious consideration, rather than the conviction that everything will stay as it is.
The AI-powered workplace of the future therefore changes more than what people do for work. It also changes how companies can recognise, assess and respond to change.
Perhaps that very ability to adapt will become one of the most important capabilities a company can have.