The City of Raisio, a municipality of around 25,000 residents in southwestern Finland, has embarked on a deliberate journey to integrate Microsoft 365 Copilot into its daily operations. In autumn 2025, the city partnered with Sogeti, part of Capgemini, to train nearly 100 municipal employees through a staged program designed to foster confident, effective use of the AI assistant across departments. This initiative reflects a growing trend among public sector organizations to embrace generative AI, but with the caution and structured change management essential for taxpayer-funded services.
Raisio’s decision to deploy Copilot did not happen overnight. Like many local governments, the city had been exploring digital tools to streamline administrative work, improve citizen services, and reduce repetitive manual tasks. When Microsoft introduced Copilot for Microsoft 365 in late 2023—embedding large language model capabilities into Word, Excel, PowerPoint, Outlook, and Teams—it promised a leap in productivity. For a moderately sized municipality, the lure of automating meeting summaries, drafting official communications, and analyzing budget data in natural language was compelling. Yet Raisio’s leadership recognized that technology alone would not guarantee success. Without proper training and cultural adaptation, even the most advanced AI can become shelfware or, worse, introduce errors that erode trust.
To mitigate these risks, Raisio turned to Sogeti, a technology and engineering services firm with deep experience in Microsoft ecosystems and change management. Sogeti’s approach, as glimpsed from the rollout, emphasizes a phased strategy. While the exact structure of the Raisio program remains undisclosed, such initiatives typically begin with a pilot group of early adopters who receive intensive, hands-on coaching. These power users then become internal champions, mentoring colleagues and providing feedback to refine training materials. Sogeti’s parent company, Capgemini, has been at the forefront of enterprise AI adoption, and its methodologies likely informed the Raisio engagement: assess current workflows, identify high-value use cases, train in context, and scale gradually while measuring impact.
The nearly 100 employees targeted in the initial phase represent a cross-section of Raisio’s workforce—administrative assistants, department heads, planners, and customer service staff. For many, the introduction of an AI copilot will be their first direct interaction with generative AI at work. The training therefore must demystify how Copilot works, set clear expectations about its capabilities and limitations, and embed responsible use principles. Municipal employees handle sensitive data, from social welfare records to urban planning documents, so data residency and compliance with EU regulations like GDPR are paramount. Sogeti’s trainers would need to address how Copilot operates within the Microsoft 365 compliance framework, ensuring that prompts and generated content remain within the tenant and are governed by existing data policies.
One of the core benefits Raisio hopes to unlock is time savings in routine documentation. For instance, a city planner might use Copilot in Word to draft a zoning amendment based on previous public comments, or a finance officer could ask Copilot in Excel to forecast revenue trends from a spreadsheet without complex formulas. In Teams, automatic meeting recaps and action items can cut the overhead of internal coordination. These improvements, multiplied across dozens of employees, could yield hundreds of hours freed for higher-value public service. Yet the training must go beyond feature demonstration. Employees need to learn how to craft effective prompts, verify AI-generated content for accuracy, and recognize when human judgment must prevail.
Raisio’s phased approach also acknowledges the psychological hurdles of AI adoption. Some staff may fear job displacement or feel overwhelmed by a tool that threatens established workflows. Sogeti’s change management experts likely incorporate techniques to build psychological safety: celebrating small wins, involving skeptical employees in feedback loops, and communicating that Copilot is an augmenter, not a replacer. This human-centric focus mirrors best practices observed in other public sector AI projects across Europe, from chatbots in Stockholm’s city services to automated document analysis in Dutch municipalities. What sets Raisio apart is the deliberate front-loading of training before scaling, a lesson drawn from early Copilot deployments where rushed rollouts led to confusion and underutilization.
The choice of autumn 2025 for this rollout is noteworthy. By then, Microsoft will have had two years to refine Copilot based on enterprise feedback, adding features like Business Chat, which allows cross-application queries, and deeper integration with Microsoft Graph. The AI landscape in the EU will have also matured, with the AI Act likely in full effect, providing a regulatory floor for public sector use. Raisio’s timing avoids the bleeding edge and piggybacks on learnings that enterprises have already absorbed. It also aligns with Finland’s broader digital strategy, which emphasizes human-centric AI and life-long learning for public employees.
On the technical front, deploying Copilot for Microsoft 365 requires specific prerequisites: Microsoft 365 E3 or E5 licenses, the Copilot add-on, and proper configuration of data classifications and sensitivity labels. Sogeti’s engagement no doubt includes a readiness assessment of Raisio’s IT environment, ensuring that the infrastructure can support AI-powered features securely. While the city may not have the scale of a multinational corporation, its IT team must still manage data governance, user permissions, and monitoring for potentially sensitive queries. Training thus covers both the front-end experience and the dos and don’ts of handling information via AI chat.
Real-world examples from other early adopters illustrate what Raisio might expect. Several European local governments have piloted Copilot with varying results. In one case, a city council in the UK reported that Copilot reduced the time to draft reports by 30%, but only after intensive training and the appointment of dedicated AI champions. Another municipality in the Netherlands saw mixed adoption rates, with younger employees embracing the tool while older staff struggled with the conversational interface. These experiences underscore the value of Sogeti’s staged program: by gradually building competence and confidence, Raisio can avoid a two-speed office where AI widens skill gaps.
Measuring the success of this training initiative will require clear metrics. Raisio and Sogeti likely defined key performance indicators upfront: user activation rates, frequency of meaningful Copilot usage, reduction in time spent on specific tasks, and employee satisfaction scores. Because municipal workflows are often less transaction-driven than those in private enterprises, qualitative feedback—such as how Copilot aided a complex social service case assessment—may be equally important. The staged rollout allows for iterative adjustment: if the first cohort reports that certain features are more valuable than others, training for subsequent groups can be tailored accordingly.
Looking ahead, Raisio’s experience could serve as a template for other Finnish municipalities and even inspire broader regional adoption. Finland has a tradition of digital collaboration between public bodies, and the lessons learned in Raisio about effective training, managing fears, and ensuring compliance could feed into national guidelines. Sogeti, with its Capgemini backing, may be well-positioned to replicate this model across the Nordic region, where trust in public institutions and a careful approach to technology adoption align well with staged AI implementations.
Yet challenges remain. Copilot, like all large language model systems, can generate plausible but incorrect information—in AI parlance, hallucinations. In a public sector context, a mistakenly invented policy reference or an inaccurate financial figure could have serious ramifications. The training must therefore ingrain a verification mindset: employees must treat Copilot as a junior assistant whose work always requires review. Sogeti’s curriculum likely includes exercises where trainees identify errors in AI-generated texts, reinforcing critical engagement.
Another consideration is equity. As Raisio scales Copilot beyond the initial nearly 100 users, it must ensure that all roles benefit, not just those in administrative or knowledge-work positions. Field workers, maintenance crews, and customer-facing staff may also benefit from AI capabilities on mobile devices, such as summarizing maintenance histories or translating documents in real time. The phased nature of adoption allows the city to gather data on which roles gain most from Copilot and plan subsequent phases to close gaps.
As the rollout progresses through autumn 2025, the eyes of Finland’s public sector IT community will be on Raisio. The city’s collaboration with Sogeti is not merely a technology project; it is an experiment in organizational learning. If successful, it will demonstrate that even smaller municipalities can harness advanced AI responsibly, improving both employee productivity and the quality of public services. For Microsoft, it provides a valuable public sector case study at a time when governments worldwide are scrutinizing AI investments. The Raisio story reinforces that the path to AI value runs through people, not just algorithms, and that the most important copilot is the one that guides users toward confident, safe, and effective adoption.