From talk to action: how leaders are executing AI transformation

Over the last 18 months, artificial intelligence, especially generative AI has made the leap from abstract hype to tangible action in Nordic organizations. From national regulators to energy disruptors to healthcare giants, leaders have moved fast to explore what generative AI might do for their business and operations. Mostly all have adopted the personal productivity tools that are easily available, Some have built early agents. Others launched proof-of-concept pilots or wide company level capability programs. A few went deeper, embedding generative AI into specific processes everyday to those essential everyday workflows.

The energy of experimentation has also run headfirst into the reality of operational complexity. Legacy systems don’t cooperate. Data is scattered. AI governance is unclear. The people responsible for scaling the solutions often lack the mandate to do so, and the people with the mandate often lack the clarity. What we are seeing in these cases is not the failure of AI technology. It is the failure of execution environments and us not being ready to lead or learn fast enough. This pattern is consistent with recent research from MIT, which found that 95% of generative AI pilots fail to deliver meaningful business value to profit & loss. Key reasoning being the above described learning gap not the technology, but the implementation, integration and culture. (MIT, The GenAI Divide: State of AI in Business 2025).

The shift we are seeing

At Futurice, we’ve spent the past months talking to Nordic executives, hosting workshops, building the solutions and analyzing both the success stories and the slowdowns. The same pattern emerges in every industry: AI initiatives will not deliver wide business impact until it is treated not as a spot on solutions but as an operating model shift.

This shift is not about adding intelligence to isolated parts of the business. It is about how to boldly reimagine how the business runs. Where decisions are made. How processes flow. What can be automated? Where human judgment adds the most value.

One public leader described this moment to us as the end of AI as an “innovation playground”. It is now becoming part of the core stack. And that transition demands a different mindset, a different architecture, and a different kind of leadership.

From the frontlines: what real progress looks like with Generative AI solutions

We’ve seen this shift playing out in very different ways. At Traficom, it looks like a regulator treating governance not as a blocker, but as an enabler of innovation using internal Generative AI pilots like AgencyBot to explore how automation can help people, not just replace them. At Kempower, a fast-scaling energy company, it means embedding AI into operations through modular platforms and clear ownership structures, where AI is aligned with growth, not siloed in experimentation.

At Terveystalo, AI transformation has meant putting the process first, understanding how patients and professionals interact, and then identifying how predictive capabilities could improve outcomes. And at ICAx, it is about applying lean experimentation at speed, with feedback loops designed for learning and scaling, not just proof-of-concept validation. These are not generic stories of AI success. They are stories of process-native transformation about changing how value is created.