In 2M-IT’s AI Project, Quality and Safety Come First

Marko Ranta serves as the Director of IT Services at 2M-IT. IT Services produces comprehensive and large-scale ICT services for wellbeing regions, ranging from basic infrastructure to telecommunications, information security, and end-user support.

“Our work at 2M-IT directly impacts how smoothly and safely critical everyday situations are managed—such as at the emergency room door, where every moment and system reliability are crucial. Seemingly minor technical glitches, like a non-functioning computer or a frozen system, can in a clinical environment even pose patient safety risks if, for example, patient information cannot be accessed at the right time. Our job is to ensure that these fundamentals work flawlessly so that professionals can focus on what matters most—helping people. This real-world impact is at the core of everything we do,” says Ranta.

The Most Valuable Role of AI in Service Production Is to Free Up Experts for More Demanding Tasks

In 2024, 2M-IT adopted the ServiceNow service management system, enabling more efficient IT service development. The system includes the AI assistant Now Assist. 2M-IT’s Support Services is currently running an AI project to develop, test, and pilot AI solutions that automate work steps and improve customer service.

“The most valuable role of AI in service production is to eliminate repetitive and manual work and clarify processes. Experts’ capacity is freed up for where the greatest value is created: high-quality problem solving and developing the customer experience. This way, we can truly focus on what benefits the customer most and continuously improve our services and cost-effectiveness,” Ranta explains.

AI assists with support requests, automatically classifies tickets, suggests solutions, and streamlines information management. AI is not rolled out to every operation at once; instead, pilots enable controlled adoption and learning.

“It has been important for us to pilot first with a smaller group and introduce AI in a controlled way. Monitoring and concrete benefits are key to safe adoption.”

Trust Is Built on Openness and Transparency

Ranta emphasizes open dialogue with customers about how AI is utilized and the added value being pursued. Trust is built through a clear governance model, risk assessment, and transparent processes—only in this way can customers be assured of high-quality and secure service.

“We need to have open discussions with customers about how AI solutions are built and how they support their operations. Trust comes from assessing risks and information security in advance and ensuring processes are transparent.”

First Steps in Automation—Three Principles

According to Marko, automation should start by following three key principles:

1. Eliminate repetitive manual work, thereby freeing up significant time for experts to handle more demanding tasks.
2. Ensure data quality—automation’s functionality and reliability are based on high-quality data.
3. Improve the customer experience and workflow: refining processes is directly reflected in making customers’ everyday lives easier.

“When we choose which processes to automate first, we pay special attention to tasks that free up expert time and where data quality enables reliable automation. Refining processes is directly reflected in the customer experience—daily work becomes easier and turnaround times improve.”

Practical Experiences: The Service Desk Transformation Is Already Underway

Marko illustrates with a concrete example: previously, specialists manually classified every customer service request—a total of half a million per year—and searched for solutions in the knowledge base. In the future, AI will handle classification and propose solutions, reducing manual work and significantly speeding up request handling and turnaround times. The goal is that by December 2027, 95% of incoming requests will be classified automatically, resolution time will be reduced by an average of 50%, and customer satisfaction will remain at its current level.

However, the adoption of AI is not just a technical project but requires reviewing the entire operating model. Marko emphasizes the importance of data quality and process functionality: “Without high-quality data, automation doesn’t work—it’s not just about technology, but about developing the whole system.”

The biggest lesson has been understanding that implementing AI requires assessing the entire process and data quality. Otherwise, automation won’t produce reliable results. It is also important to ensure that guidelines and documentation practices support AI operations.

Metrics and Safety—Measuring Success

The success of the automation project is measured by, among other things, shortening ticket queues and resolution times, as well as customer satisfaction. In addition, processes, responsibility matrices, and documentation guidelines have been updated to meet AI requirements—this has enabled the construction of a reliable, high-quality, and traceable solution.

“The most important metrics are shorter ticket queues, faster resolution times, and increased customer satisfaction. It’s also essential that our experts adopt the new operating models and that processes support automation. Clarifying processes and responsibility roles has been crucial for quality and safety.”

Transferring Routines to AI—Where Are We Now?

At 2M-IT, AI already classifies tickets, produces summaries, suggests solutions, and assists in drafting knowledge articles as well as transcribing calls. This has reduced manual work and improved service speed and consistency. In the future, AI capabilities will expand to more teams and processes.

“We are already well along in having AI handle manual tasks—ticket classification, summary creation, and drafting knowledge articles have become automated. The next goal is to expand these capabilities to more teams and everyday processes.”

Scaling, Competence, and Partnerships—The Next Steps

In the next six months, the goal is to scale successful AI solutions even more broadly. This requires strong partnerships and increased expertise in both technology and governance models. Marko sees that success will be measured by whether AI truly reduces person-years and genuinely boosts efficiency—numbers and practical everyday realities will tell.

“We don’t yet have all the expertise in-house, so partnerships are essential, especially in the early stages of implementation. At the same time, we are growing our own skills and capabilities in both technology and governance. The most important thing is that AI truly reduces person-years and brings real efficiency to daily operations.”

When Is the AI Project Finished?

“Perfect readiness may never be achieved, development is always ongoing. However, when pilot results, risks, and metrics show that goals have been met and the organization’s capabilities are at the right level, we can move to production,” Marko summarizes.

If the goals are not met, the project will not move to production. The line is clear and honest: quality and safety are uncompromisingly the top priority.

“It’s hard to say when an AI project is fully complete, development is continuous. When the criteria, risks, and metrics set in the pilot are met and resources are in place, we can move to production. If the goals are not achieved, the project will not proceed—quality and safety take precedence over everything else.”

Photo: Pasi Leino Photography

You might be interested in learning more:

Artificial intelligence enhances project quality and ensures continuous learning

Intelligent Service Delivery at 2M‑IT – ServiceNow and the Opportunities of AI in Support Services