This episode took me into a part of the AI conversation that gets less airtime than it deserves: the public sector. My guest was Nina Nissilä, Director and 2nd Deputy Director General at Kela, Finland’s Social Insurance Institution.
A seven-year modernisation, 8,000 people deep
Kela is partway through EEPOS, one of Finland’s largest IT modernisation programmes, rebuilding the systems behind benefits processing for the country’s 5.5 million residents. Nina leads that as a strategic effort, not a technical one, and the change-management challenge of moving 8,000 people through it was one of the most grounded parts of the conversation.
Where AI earns its place, and where it doesn’t
The line I found most useful from Nina: not every efficiency problem is an AI problem. A lot of what Kela needed was well-scoped, rule-based automation, reliable and auditable by design. AI gets introduced where it earns its place on top of that foundation, moving from early experiments toward real, production use cases.
Trust is the constraint that shapes everything
Public trust, data handling, and responsible AI are not side considerations for an institution like Kela. They are the constraint the whole roadmap is designed around, out to 2035. We also talked about how Finland’s position in EU AI policy shapes that roadmap, and where public-sector AI adoption genuinely differs from the private sector rather than simply lagging it.
Finland is often called Europe’s most digitally advanced nation. This conversation is a good look at why, and at what “digital-and-AI-assisted public service” actually takes to deliver at national scale.