The development and utilisation of artificial intelligence continues at an accelerating pace. What kind of effects will it have on business when AI agents become more common and their use becomes more diverse? On the other hand, regulation and the international operating environment are shaping the use of AI. The rise of robots is also beginning.
Digia's AI experts, Chief Technology Officer Juhana Juppo and Senior Vice President Sami Paihonen, discuss the most significant trends for next year and how they will impact the operations of organisations and the development of competitiveness.
Last year, the most significant change in the use of artificial intelligence in companies was the introduction of AI agents.
However, there is still a lot to draw from. Gartner, for example, has predicted that next year, 40 per cent of companies will deploy AI agents, compared to only 5 per cent this year. So the change is really wild.
"We are still quite early with the agents. Next year, we can expect an increase in automation, which will mean that agents will become even more independent. It frees up people's working time for more suitable jobs," says Chief Technology Officer Juhana Juppo.
Similarly, the use of different agent teams is becoming more common. You no longer use just one agent to do a straightforward task. Instead, a group of agents can be created who work together to solve different situations, especially in knowledge work.
"One person investigates the matter, another writes about it, a third checks and corrects, and a fourth coordinates the entire project. Such a team of agents is sure to achieve the best outcome."
As the use of agents evolves, people's job descriptions will also undergo significant changes.
"We are starting to see pairs or teams where the AI agent is one independent member among people. It is not just about a search robot, for example, but the agent brings genuine new creative added value to the team's operations," says Senior Vice President Sami Paihonen.
Often, new and revolutionary technology is first used very simply, for example, by utilising only one capability.
The same has happened with extensive language models. However, there is now a change in the way we use them.
"In the case of AI, too, we are starting to move more and more towards accuracy, and we are no longer resorting to brute force alone. It is also a matter of necessity, because if AI is used recklessly, we will run out of electricity," Paihonen says.
The end of electricity is even a bit sad, considering that extensive language models have only been in use for three years.
However, according to forecasts, the energy use of data centers, the lifeline of AI, will double by the end of this decade. It will inevitably force us to use AI more economically.
"Hybrid use is becoming more common. We have local, regional and global models, and they are all used as needed," Juppo says.
In other words, the AI competition is also a competition for energy. Countries with stable, clean, and affordable energy are strong in this regard. Finland is one of them, which is why data center investments evoke strong emotions.
"Finland has the capability and know-how to build and operate data centers. However, I would not want Finland to remain Europe's AI machine room, but we should move up the value chain," Paihonen says.
In recent years, the EU has introduced numerous regulations related to AI, and this is increasingly being reflected in business operations.
"AI developers and solution providers need to think more carefully than ever to ensure that regulatory compliance is included in all deliveries," Juppo describes.
For companies operating extensively in Europe, the problem may be that, although the law is written in the same way for everyone, different countries interpret it in their own way.
"The Italian authority monitors the implementation of the Data Act differently from its Finnish counterpart. This may become a kind of domestic political problem and a problem affecting competitiveness in the EU area," says Sami Paihonen.
"We see big leaps internationally, but also nationally. For example, the Finnish Defence Forces is investing heavily in AI and has announced that it will open an artificial intelligence center together with partners at the beginning of next year," says Juhana Juppo.
The use of robots is also becoming more common in the defence and security sector.
"If the cloud is down when it's crunching, the computing goes to robots equipped with AI on the battlefield. This way, the ability to function is maintained for longer," Paihonen says.
Until now, AI has mainly been considered one of the areas of IT. However, rapid development is also taking place in so-called physical artificial intelligence – i.e. robotics. The combination of humanoid robots and extensive language models is becoming a very efficient manual labourer.
"The development of humanoid robots has been completely incomprehensible lately. Until now, their use has primarily been demonstrated in production environments; however, there are already versions being deployed in production use. It has a strong impact on performance," Juppo says.
He points out that production outsourced to countries with cheaper labour can be returned back to Finland. The use of robots is more or less the same everywhere in the world, because the cost of work consists mainly of maintenance and the price of electricity.
In 2026, we will see the first large-scale deployments of humanoid robots, with a more significant impact expected to occur between 2027 and 2028.
A year ago, Juhana Juppo and Sami Paihonen predicted that AI agents would become more common at a rapid pace, and this has indeed happened (article in Finnish). They also indicated that the number of AI models will increase and "sustainable AI" will become a trend. Especially in terms of energy, there is still room for improvement in the sustainability of artificial intelligence.
The third prediction concerned the impact of AI on the creative industries. There has been no significant turmoil yet, but it is likely to begin next year.
Fourthly, Juppo and Paihonen estimate that industry-specific AI killer applications will emerge in various industries, disrupting current operating methods. They may be expected even more extensively.
The fifth forecast concerned the current turbulent social situation. AI is also a tool for criminals and scammers, and its malicious use is on the rise. In that, Juppo and Paihonen were exactly right.
The benefits enabled by AI can only be achieved when it is available for production use and as part of the development of operations, safely.