Four paths to AI success – why some companies are falling behind

A recent survey of 1,250 employees across the Nordics reveals that organisations in Denmark, Norway, and Sweden are actively adopting AI. In Denmark, 56% of organisations offer AI tools to employees, in Norway 75%, and in Sweden 54%. And employees are generally positive: over 80% see AI as helpful for their daily work.

However, there are major differences in the benefits companies gain from using AI. There’s a clear gap between the top and bottom performers, and it can’t be explained by industry, size, or other traditional factors. To understand it, we need to look at how AI is being used. Here, four archetypes of usage patterns emerge, each representing a different approach to AI.

The first archetype uses AI as a “task assistant.” This involves using AI to answer questions or generate content like emails or meeting summaries. It typically relies on simple prompting of language models like Copilot, ChatGPT, or Gemini. This is the most basic level and requires minimal implementation.

The next level, the “knowledge agent,” focuses on a specific area of expertise. Examples include self-service support, where employees can get help with IT issues or customers can get help with their order. This requires more effort, as the right data sources must be identified and orchestrated.

The third type, the “business problem solver,” tackles more complex challenges such as forecasting, optimisation, or classification. Examples include supply chain optimization or predictive maintenance. This level requires tailored models and data, as well as changes to business processes.

“Autonomous agents” represent the most advanced form of AI. These systems make decisions and act without human intervention. This includes physical systems like self-driving cars and factory robots, and increasingly, digital solutions.

In the digital realm, we talk about “agentic AI”, software that completes tasks independently, without asking for permission or help. This can range from automated customer service to AI systems that plan and execute complex business operations.

Using agentic AI requires a fundamental rethink of workflows and governance. Companies must address technical, legal, and ethical risks and ensure there is clarity around responsibility, safety, and control.

The study shows a clear link between the different AI archetypes and efficiency gains. Organisations that only use AI as a “task assistant” see the least benefit. Those that have implemented agentic AI realise the greatest value. Each level brings increased value, but also increased complexity.

It’s easy to give employees access to Copilot or ChatGPT, but that alone won’t deliver the full impact. Many organisations stall when considering more advanced use cases because these require greater investment, higher risk tolerance, and significant change management.

To realise AI’s full potential in the Nordics, access to tools isn’t enough. It requires a focused, holistic effort that involves implementation, rollout, governance, and adaptation of the operating model.

Read the article in Danish Finans.