An agent takes care of your affairs

Artificial intelligence is everywhere now. Yet surprisingly little is written or discussed about the impact of AI developments on public services, their delivery, or what needs to be done to better prepare for this transformation.

Understanding the potential of AI is difficult because of the breadth of the issue. There is no single truth or correct answer to what AI means for public service tasks, because change can mean changes in work, services, or technology. In this article, I focus on the impacts of AI on digital services in public administration.

From efficiency to prediction

In the first phase, AI was mainly applied to make existing processes more efficient. This was not a success story, as in many cases automation or robotics carried out the same tasks better and at lower cost. AI has also been used to update and develop our current digital services, for example by introducing AI-based search engines or chat functions to make it easier to find information or complete tasks on websites or intranets.

Next came the wave of generative AI and solutions for streamlining expert work. In Tampere, we introduced MS Copilot in 2024 and closely monitored the benefits reported by experts. The results were staggering; the ROI on the investment was over 1,000% in multiple metrics. The benefits stemmed from the time saved by experts.

At present, we are working with scalable AI models. AI is highly capable of making data-driven predictions, and we currently use it, for example, to forecast traffic volumes in the city centre and visitor numbers in museums. Work is under way to apply the forecasting model to areas such as predicting the number of learners and optimising food orders based on those forecasts. The focus has therefore moved beyond saving working time alone to a new phase where benefits can be measured not only in time saved but also in reduced costs of physical resources.

Personalised service through a single channel

The real turning point for AI will come when we build public digital services to be AI-native. This means an agent-based architecture in which an autonomous AI agent receives a user’s request and carries it out. The request might be something like arranging the practical matters related to a family’s move. The agent would then find housing in the new locality, arrange day care and school places for the children, identify possible social benefits and update the tax rate according to the salary in the new job. The user could refine the agent’s suggestions or approve them.

Artificial intelligence finally provides the tools to implement a one-stop solution for public services. It removes the need to deal separately with different authorities’ services, as the entire process can be handled by conversing with a familiar AI agent. This could be done either in writing or by speaking, meaning that technological development may enable a return to personalised service. Instead of another person, the counterpart would be a machine that understands all languages and carries out tasks on our behalf. In the future, the ability to use a computer will no longer be a prerequisite for accessing digital services.

Change is inevitable

Wild speculation and dangerous daydreaming? Autonomous AI agents are already solutions in production use. So easy and convenient that they will take over the service market in no time.

The more relevant question is whether the public sector is ready for the change. What should we do when agents start knocking on the door of our services, using the Suomi.fi identification authorised by their users?

We need a next-generation agent architecture for public services. A concept that ensures the interoperability, security, data protection, and identity management of AI-based public administration services. We must prepare for the fact that applications like those we have today will change radically and largely disappear, as agents will handle the same tasks on behalf of people.

Interoperability can only be achieved through collaboration, something no single actor can design alone That is why it is important to establish a shared vision and concept that each government agency can implement within its own area of responsibility.

AI poses challenges, but it also promotes equality. That is why we should act now, while it is still possible, rather than waiting for the moment when action becomes unavoidable.