How do you turn artificial intelligence into a real working tool in a municipality with thousands of employees, high complexity, sensitive data and a major responsibility towards citizens?
In Gentofte Municipality, they have chosen an approach that is both ambitious and down-to-earth: AI must be safe, useful and anchored with the people who know the work best. Not only in IT. Not only in management. But also among frontline staff and citizens.
That is what Isabel Vittrup-Pallier, AI specialist in Gentofte Municipality, explained in the podcast EDB 5.0, where she described how Gentofte works to make AI accessible in practice. Her point is simple but important: when new technology is meant to create real value in the public sector, it must be developed close to the people who work with the tasks every day.
Gentofte Municipality was early in taking AI seriously. In just a few years, the municipality has built a setup where technology, governance, competence development and citizen involvement are connected.
Isabel Vittrup-Pallier is an AI specialist in Gentofte Municipality and works to spread, procure, implement and support the use of AI tools across the organisation. But her path into the role is at least as interesting as the role itself.
She has a background from the IT University of Copenhagen with a focus on UX, change management and technology in large organisations, and she actively chose the public sector. Not because it is the easiest arena to work in, but because this is where technology can make a difference close to people.
As she describes it, the motivation is to support the people already doing the most important work:
"In the public sector, I can at least be a bit in the background and try to help some of the people who actually make a difference have a better and more enjoyable working life."
That perspective runs through all of Gentofte's AI work.
Already during her thesis, Isabel worked critically with AI in healthcare. Among other things, she examined a project using artificial intelligence in a maternity ward and questioned the way the technology was discussed.
She describes the early debate as "techno-heroic", a way of talking about new technology as if it can almost automatically save the world. She has brought that critical approach into the municipality: AI should not be put on a pedestal. It should be assessed concretely, responsibly and in relation to the reality employees and citizens actually face.
That is exactly why Gentofte Municipality's approach is interesting. AI has not become a prestige project. It has become organisational work.
According to Isabel, one of the most important reasons Gentofte got started quickly was clear support from the top.
The municipality had already done substantial groundwork before she arrived: potential analyses, proofs of concept and political interest. That meant she entered an organisation that was ready to move.
According to her, it has been crucial that the direction was clear from the beginning:
"Making it very clear from the start, top-down, that this is something we want you to spend time on, has been absolutely essential."
That creates something important in everyday work. Employees do not meet AI with folded arms, but with curiosity. Not because everything is settled, but because there is legitimacy around the work.
A special part of Gentofte's work is the task committee on artificial intelligence that the municipality has established. Here, municipal politicians and citizens sit together to discuss what Gentofte Municipality actually wants with AI.
The committee consists of five politicians and ten citizens, selected with attention to both diversity and expertise. Age, gender, legal insight, experience from companies working with AI and technical understanding have all been considered.
The purpose is not only to get input. The purpose is to create a foundation for ambitions, ethics and concrete priorities.
As Isabel describes it, it is about finding the balance between courage and responsibility:
"We need to be brave. We need to be technology-ready. Of course, we need to act sensibly and ethically, but we need to dare to be very ambitious about how we use much more AI in Gentofte Municipality."
That is an important point. If AI in the public sector becomes only a question of systems and suppliers, something crucial is lost: trust. And trust requires both transparency and involvement.
Another important part of Gentofte's AI work is platform choice.
Before Isabel arrived, the municipality had already tried developing its own solutions from scratch. But it quickly proved heavy, time-consuming and difficult to scale. When employees who are not hired as developers have to build advanced solutions themselves, it takes time away from the core task.
That is why Gentofte chose another path.
As Isabel explains in the podcast:
"There were these guys who have a company called Promte, which offers an AI platform they have built. In reality, it is ChatGPT in another user interface, and then we can choose ourselves where we want it hosted. The whole data-security aspect is completely in order."
For Gentofte, it was crucial that AI could be made available in an environment where data security, compliance and control were in place.
And that was exactly where Promte fitted in.
When municipalities work with AI, data security is not something you can take lightly. It is a basic prerequisite.
Isabel says it very directly:
"You cannot make data processing agreements with ChatGPT, with OpenAI. There are too many uncertainties in relation to how they process our data and things like that, which you cannot keep track of in the same way. And that is just extremely important when you are a public authority."
That is exactly the problem many public organisations face: they can see the potential in generative AI, but they cannot simply let employees loose in tools without management, transparency and clear frameworks.
Promte is built for precisely that reality. It makes it possible to use powerful language models within a framework where the organisation can control setup, access, governance and operations.
One of the most interesting moves in Gentofte is that AI is not reserved for a small specialist group.
The municipality has around 6,000-7,000 employees, and the ambition is for AI to become a tool that can be used broadly. That includes both general chat-based AI and more specialised solutions.
On Promte, Gentofte has made it possible to work with both general chat and custom-built workspaces. At the same time, the municipality has created a structure where specially trained employees can become creators and create specialised chatbots for their own professional areas.
Around 80 employees have completed an introduction programme where they are equipped to understand guidelines, responsibility and legislation. It is not an elite technical education. It is a practice-oriented setup that makes decentralised development possible without losing control.
As Isabel says:
"We would rather give them the frameworks and rules and say: this and this is what you may and may not do. But you are the specialists in your field."
It is an approach we recognise among many of the organisations that succeed best: technology should not be owned centrally alone. It should be usable locally by the people who know the workflows.
One of the most striking things in Isabel's story is how much value emerges when professional areas are given room to develop themselves.
Instead of letting IT become the bottleneck for every idea, Gentofte has chosen to create frameworks and then give domain specialists substantial responsibility. That has led to use cases that IT would never have come up with on its own.
This includes documentation, note writing and translation.
Isabel points out that many employees with citizen-facing contact spend a disproportionate amount of time documenting conversations, writing minutes or cleaning up notes. That applies, for example, to healthcare professionals, case workers and others who first visit a citizen and then return to document everything correctly.
Here, AI can make a real difference. Not by taking over decisions, but by supporting the work around them.
She also mentions translation as a concrete area where the need is large. Google Translate is effective, but not necessarily usable in a municipal context. That is why it is interesting when municipalities share solutions with each other.
As an example, she highlights Kalundborg Municipality:
"Kalundborg Municipality has spent its resources developing a municipal translation solution that can be integrated into the Promte interface. So it is something our users already know."
It shows something important: when the platform is flexible, it becomes possible to build further across organisations.
Another strong point from the episode is the need for collaboration between municipalities.
When Isabel started, there were few others in similar roles. Today the picture is different. More municipalities have built networks, and KL plays a role in bringing people together, sharing experience and creating professional forums.
According to Isabel, there is a growing recognition that municipalities need to share more with each other:
"We cannot reinvent the wheel 98 times. We need to talk to each other, and we need to collaborate and share."
That applies to experience, guidelines, solutions and considerations. Municipalities work under many of the same legal frameworks and broadly solve the same types of tasks. That is why it makes sense to reuse more and build on what already works.
In the public debate, AI is often described as a tool for saving full-time equivalents. Isabel challenges that logic.
She points out that the reality in municipalities is more often characterised by a shortage of hands, especially in citizen-facing roles. It becomes too simplistic when AI is discussed only in terms of savings.
Instead, she calls for a different conversation: how can AI improve working life for the people closest to citizens?
"Why can't we talk about how we can help them better and increase their sense of quality and joy in their work?"
That is an important shift. When AI is used well, it is not only about efficiency. It is also about removing pseudo-work, reducing friction and freeing up time for what employees are actually there to do.
Although Gentofte has come far, the work is far from finished. The next major step is anchoring.
The municipality has already identified employees who are both professionally strong and good at developing solutions. Now the task is to make them ambassadors, so they can also help colleagues understand and use AI in practice.
Isabel explains that the technical part is not necessarily the hardest. The difficult part can just as easily be standing up in front of colleagues and communicating what makes sense and what to be aware of.
That is why ambassador programmes and local anchoring are a natural next phase.
If there is one recurring principle in Gentofte's approach, it is this: the best AI solutions are not created furthest away from reality, but closest to it.
That applies to citizens, frontline staff and professionals. They know the consequences. They know where the work is under pressure. And they are best placed to assess where AI makes sense and where it does not.
That is also why Gentofte is an interesting example. Not because they have found one magical solution. But because they have chosen a model where security, governance, citizen involvement and local innovation work together.
That is precisely the kind of approach we believe is needed if AI is to create real value in the public sector.
At Promte, we help municipalities and organisations adopt AI in a way that fits reality: with data security, compliance and governance under control, and with room for employees to create value close to their own workflows.
If you want to hear more about how Promte can be used in your organisation, get in touch with us. We are happy to have a non-binding conversation and show how you can get started with AI in a way that fits your needs, your employees and your security requirements.
