The City of Raisio in Finland began a Microsoft 365 Copilot adoption program in autumn 2025 with Sogeti, part of Capgemini, training nearly 100 municipal employees ahead of a broader 2026 push to make data and generative AI part of daily public-sector work. The interesting part is not that another organization bought Copilot licenses. It is that Raisio treated the purchase as the least important part of the story. In an AI market still addicted to launch announcements, the city’s slow, human-centered rollout is a useful reminder that workplace AI succeeds or fails at the level of trust, habit, and governance.
The dominant enterprise AI story of the last two years has been speed. Vendors pitch acceleration, executives demand pilots, and IT departments are asked to turn sprawling estates of email, documents, meetings, Teams chats, SharePoint sites, and security permissions into something an assistant can safely reason over. The promise is seductive: less searching, faster drafting, smarter summarization, and fewer hours lost to administrative sludge.
Raisio’s project starts from a different premise. The city did not frame AI as a technological inevitability that employees should simply absorb. It framed AI adoption as a workplace learning problem, which is both less glamorous and far more realistic.
That distinction matters because municipal work is not a neat productivity sandbox. A city government touches education, infrastructure, HR, communications, resident services, business services, administration, and politically sensitive decision-making. The same organization may hold routine meeting notes, personnel data, planning documents, service records, citizen communications, and internal policy drafts. Generative AI does not merely add a new button to Word or Outlook; it changes how employees retrieve, compose, summarize, and possibly expose information.
Raisio’s decision to put people first is therefore not soft change-management language. It is risk management. If employees do not understand what Copilot is good at, where it can mislead, and how it relates to the city’s data, the organization has not adopted AI. It has only distributed licenses.
Raisio’s leadership appears to have understood that the expensive part of Copilot is not only the subscription. The real cost sits in the operational changes required to make the tool useful: training, governance, permissions hygiene, scenario design, employee confidence, and sustained practice. For IT leaders, that is the uncomfortable truth behind almost every enterprise AI deployment. You can buy access centrally, but you cannot buy fluency the same way.
Microsoft 365 Copilot is particularly dependent on organizational context. Its value comes from working inside the Microsoft 365 environment, where employees already live in Outlook, Teams, Word, PowerPoint, Excel, OneDrive, and SharePoint. That gives it reach. It also makes the consequences of sloppy rollout more serious, because Copilot’s usefulness depends heavily on the quality, structure, and access controls of the underlying data estate.
A city hall is exactly the kind of place where the gap between theoretical productivity and practical deployment becomes visible. A communications worker may want help drafting a resident update. An HR specialist may need to summarize policy material without leaking sensitive personnel context. An education administrator may want to turn meeting notes into actions. An infrastructure team may use AI to organize technical documentation. Those use cases sit under one municipal brand, but they are not the same job.
That is why Raisio’s staged learning model is more important than the licensing news. The city did not assume generic AI training would work for a multisector public organization. It built a sequence: broad introductory training, role-specific workshops, and a first look at automation. That is the difference between “here is a chatbot” and “here is how your work may change.”
That makes it a useful test case for Microsoft 365 Copilot adoption in the real world. Municipal employees are not all knowledge workers in the same sense. Some spend their days in documents and inboxes. Others operate closer to service delivery, planning, citizen interaction, or field coordination. Their tolerance for AI experimentation will vary, as will the amount of time they can realistically devote to training.
The city’s own framing emphasizes that its strategy elevated data as an enabler of everyday work and decision-making. That language sounds familiar because almost every public-sector digital strategy now says something similar. The harder question is how data becomes useful without turning employees into reluctant data analysts or exposing them to tools they do not trust.
Copilot offers one possible answer: put generative AI into the productivity applications employees already use. But embedded AI can also create the illusion of simplicity. If a tool appears inside Word or Teams, employees may assume it is just another feature, rather than a new way of querying organizational knowledge. Raisio’s training-first approach acknowledges that the interface may be simple while the institutional implications are not.
There is also a cultural dimension. Public administration has good reasons to be cautious. It operates under legal obligations, political scrutiny, budget constraints, and high expectations for fairness and transparency. When a private company experiments with AI and saves a few hours on slide decks, the risk may be manageable. When a city uses AI in the flow of public service, confidence and accountability matter more.
This is where the Copilot economy becomes less about AI models and more about organizational consulting. The software is standardized, but the deployment is not. A municipality, a law firm, a hospital, and a manufacturer may all buy Microsoft 365 Copilot, yet each has different data risks, workflows, language needs, compliance obligations, and employee anxieties. The partner’s job is to translate a general-purpose assistant into credible local practice.
The Raisio program did that through assessment before instruction. Sogeti examined the needs of different employee groups before designing the learning path. That detail sounds ordinary, but it is precisely what many AI rollouts skip. Too many deployments begin with tool demos and end with a vague hope that users will “find use cases.”
Role-specific workshops are the antidote to that. Employees learn faster when the examples resemble their actual work. A prompt-writing lesson about marketing copy will not help a municipal employee understand how to summarize a council document, draft an internal memo, or extract action items from a Teams meeting. AI training becomes credible when it respects professional context.
The partnership also produced AI usage guidelines for the city. That may prove more durable than any single workshop. Guidelines give employees a shared language for what is acceptable, what requires caution, and where privacy and security boundaries sit. In an environment where unofficial AI tools are only a browser tab away, internal rules can reduce both fear and improvisation.
Pre- and post-project surveys reportedly showed that training reduced hesitation and helped employees understand both the potential and limitations of AI. That last phrase is important. A successful Copilot rollout does not convince everyone that the tool is magic. It teaches them when the tool is helpful, when it is unreliable, and when human judgment must remain firmly in control.
This is especially important for generative AI because the failure mode is not always obvious. Traditional software usually fails visibly: an error message, a crash, a missing field. AI can fail fluently. It can summarize with misplaced emphasis, draft with unwarranted confidence, or overlook context that a human would catch. Employees who are trained only to be impressed are poorly prepared for that reality.
Raisio’s emphasis on a safe learning environment is therefore more than motivational language. People need permission to experiment, ask basic questions, and compare results without embarrassment. The participant who described starting “from zero” and later using Copilot as an everyday assistant captures the point: adoption is not a binary switch. It is a gradual movement from unfamiliarity to routine use.
That is also why peer learning matters. In workplace AI, the most persuasive training may come from a colleague who has discovered a practical use case. Official training explains the tool; peer examples normalize it. When employees share what worked, what failed, and what saved them time, AI stops being an executive initiative and becomes part of the organization’s practical vocabulary.
But the deeper productivity prize is not simply faster writing. It is reducing the cognitive tax of navigating fragmented information. Public-sector employees often spend time locating the right version of a document, reconstructing decisions from email threads, turning meetings into actions, or converting policy material into usable language. AI can help with that, provided the underlying information is accessible, permissioned, and reasonably well managed.
This is where municipal AI adoption intersects with a much older IT problem: information architecture. Copilot cannot magically fix inconsistent document storage, obsolete files, weak metadata, or overly broad permissions. In some cases, it will expose those weaknesses faster. A tool that can retrieve and synthesize information across Microsoft 365 is only as safe and useful as the environment it is allowed to see.
That reality should temper the enthusiasm around early success stories. Nearly 100 trained employees is meaningful for Raisio, but it is not the same thing as full organizational transformation. The training program creates a foundation. The harder work comes later, when the city must decide which processes deserve deeper automation, which datasets need better governance, and which AI uses are inappropriate for public administration.
Still, early routines matter. If employees learn to use Copilot for low-risk tasks, they build judgment before the organization moves into more consequential workflows. Summarizing meeting notes, drafting first-pass documents, and organizing communications are not trivial tasks, but they are a safer starting point than automated decision support or resident-facing AI services. Raisio’s gradualism is a feature, not a lack of ambition.
Governance in this context is not a committee that says no. It is the set of conditions that lets employees say yes safely. Workers need to know whether they can paste certain information into prompts, when they should verify outputs, how to treat AI-generated text, and whether Copilot’s access reflects existing permissions. They also need clarity on what kinds of tasks remain human-owned.
For a city, that clarity is essential. Municipal government must maintain public trust. If employees use AI to improve internal efficiency, residents may welcome the benefits. If AI appears to influence decisions without transparency, or if sensitive information is mishandled, trust can evaporate quickly. The line between internal productivity and public accountability must be deliberately drawn.
Microsoft’s enterprise pitch for Copilot leans heavily on security and integration with Microsoft 365 controls. That matters, but it does not remove the need for local governance. A technically secure platform can still be used unwisely. A permissions model can still reflect years of accumulated access sprawl. A user can still overtrust a confident summary.
Raisio’s co-created AI guidelines suggest a better model: make policy part of adoption rather than an after-the-fact restriction. Employees are more likely to follow rules they understand in the context of their own work. The point is not to bury AI under bureaucracy. It is to make the boundaries visible enough that experimentation can continue without reckless improvisation.
Raisio’s CIO framed the employer’s role as enabling people to succeed, making work smoother, and supporting well-being. That is a notable shift in emphasis. The value proposition is not only “do more with less,” the phrase that haunts public-sector technology projects. It is also “reduce unnecessary friction so employees can focus on meaningful work.”
That framing is politically and operationally smarter. Workers are understandably wary when AI is introduced as a productivity mandate. If the message is that AI exists to squeeze more output from the same headcount, employees may comply without trust. If the message is that AI can remove routine burdens while preserving professional judgment, the organization has a better chance of honest adoption.
But that promise must be earned. Training cannot be a one-off webinar. It must account for new features, changing policies, emerging use cases, and employee feedback. Microsoft is evolving Copilot rapidly, and the line between assistant, agent, automation layer, and workflow orchestrator will keep shifting. IT and digital teams will need to help employees keep up without turning every update into a new anxiety cycle.
In that sense, Raisio’s open access to materials after the project matters. Self-study resources allow learning to continue beyond scheduled sessions. They also help employees revisit concepts when they encounter a real task, which is often when training finally becomes relevant.
This is where the risk profile changes. A human asking Copilot to summarize a document can review the output before using it. An automated workflow that routes information, drafts responses, updates records, or triggers actions requires stronger controls. The move from assistant to agent raises the stakes.
For municipalities, that transition must be handled carefully. Automating internal administrative steps may deliver real benefits. Automating anything that affects residents, entitlements, inspections, services, or official decisions requires a much higher bar. Even if AI is not making the final decision, its role in shaping the information available to decision-makers deserves scrutiny.
Raisio’s gradual path gives it a better chance of navigating that future. Employees who understand AI’s limits are more likely to design sensible automation. Leaders who have listened to staff concerns are more likely to spot where process change is needed before technology is added. IT teams that have built guidelines early are better positioned to extend them into more complex scenarios.
The danger for every organization is the temptation to treat automation as an inevitable next step. It is not. Some workflows should be accelerated, some should be redesigned, and some should remain deliberately human. The point of AI maturity is not to automate everything. It is to know the difference.
On the opportunity side, Copilot sits in software many organizations already use. That gives Microsoft a distribution advantage few competitors can match. If employees live in Microsoft 365 all day, an AI assistant embedded there has a natural path into everyday work. For public-sector organizations already standardized on Microsoft tools, Copilot may feel like the least disruptive AI option.
On the problem side, that same familiarity can encourage underinvestment in rollout. Because Copilot appears inside known applications, leaders may underestimate the need for training and governance. They may assume employees will simply learn by doing. Some will, but many will not, and the unevenness can create disappointing usage metrics, inconsistent practices, or quiet workarounds using unsanctioned AI tools.
Microsoft therefore needs stories like Raisio’s. They demonstrate that Copilot adoption can work when paired with structured learning, partner support, and executive sponsorship. They also subtly shift responsibility back to the customer: if the tool disappoints, did the organization prepare its data, train its people, and define its policies?
That is fair only up to a point. Vendors should not get to sell transformation and then blame customers for needing transformation work. But the practical lesson remains: Copilot is not self-implementing. The organizations that get value will be the ones that treat adoption as a program, not a toggle.
That modesty is precisely why the case is worth attention. Most organizations do not need another breathless claim that AI will reinvent work overnight. They need examples of how to begin without breaking trust. Raisio’s project offers a pattern: align AI with strategy, assess employee needs, train in stages, build guidelines, allow self-paced learning, and keep communication open.
The city’s leadership also avoided a common mistake: separating technology from culture. It chose a partner partly because technical expertise and change management were treated as inseparable. That should be obvious by now, but in enterprise IT it still is not. Too many projects divide the world into deployment on one side and adoption on the other, as though users are an afterthought.
The better view is that adoption is the product. A Copilot license sitting unused, misused, or feared is not a productivity tool. It is a recurring cost. A Copilot license in the hands of an employee who understands when to use it, how to verify it, and what not to feed it is something different: a small but compounding change in how knowledge work gets done.
Raisio’s path also has the advantage of being replicable. Not every city has the same budget, partner access, or Microsoft footprint, but the principles travel well. Start with work, not features. Train by role, not slogan. Publish rules early. Encourage peer learning. Measure confidence as well as usage.
Raisio Makes the Boring Part the Breakthrough
The dominant enterprise AI story of the last two years has been speed. Vendors pitch acceleration, executives demand pilots, and IT departments are asked to turn sprawling estates of email, documents, meetings, Teams chats, SharePoint sites, and security permissions into something an assistant can safely reason over. The promise is seductive: less searching, faster drafting, smarter summarization, and fewer hours lost to administrative sludge.Raisio’s project starts from a different premise. The city did not frame AI as a technological inevitability that employees should simply absorb. It framed AI adoption as a workplace learning problem, which is both less glamorous and far more realistic.
That distinction matters because municipal work is not a neat productivity sandbox. A city government touches education, infrastructure, HR, communications, resident services, business services, administration, and politically sensitive decision-making. The same organization may hold routine meeting notes, personnel data, planning documents, service records, citizen communications, and internal policy drafts. Generative AI does not merely add a new button to Word or Outlook; it changes how employees retrieve, compose, summarize, and possibly expose information.
Raisio’s decision to put people first is therefore not soft change-management language. It is risk management. If employees do not understand what Copilot is good at, where it can mislead, and how it relates to the city’s data, the organization has not adopted AI. It has only distributed licenses.
The Copilot License Was the Starting Gun, Not the Finish Line
One of the most telling details in the Raisio story is that Microsoft 365 Copilot licenses had already been purchased before the training program took shape. In many organizations, that would have been enough to declare progress. Procurement is measurable. Adoption is messier.Raisio’s leadership appears to have understood that the expensive part of Copilot is not only the subscription. The real cost sits in the operational changes required to make the tool useful: training, governance, permissions hygiene, scenario design, employee confidence, and sustained practice. For IT leaders, that is the uncomfortable truth behind almost every enterprise AI deployment. You can buy access centrally, but you cannot buy fluency the same way.
Microsoft 365 Copilot is particularly dependent on organizational context. Its value comes from working inside the Microsoft 365 environment, where employees already live in Outlook, Teams, Word, PowerPoint, Excel, OneDrive, and SharePoint. That gives it reach. It also makes the consequences of sloppy rollout more serious, because Copilot’s usefulness depends heavily on the quality, structure, and access controls of the underlying data estate.
A city hall is exactly the kind of place where the gap between theoretical productivity and practical deployment becomes visible. A communications worker may want help drafting a resident update. An HR specialist may need to summarize policy material without leaking sensitive personnel context. An education administrator may want to turn meeting notes into actions. An infrastructure team may use AI to organize technical documentation. Those use cases sit under one municipal brand, but they are not the same job.
That is why Raisio’s staged learning model is more important than the licensing news. The city did not assume generic AI training would work for a multisector public organization. It built a sequence: broad introductory training, role-specific workshops, and a first look at automation. That is the difference between “here is a chatbot” and “here is how your work may change.”
Finnish Local Government Is a Stress Test for Practical AI
The Raisio case is small enough to be concrete and broad enough to be instructive. A Finnish city is not a hyperscale cloud company, a bank with a massive compliance department, or a startup with ten employees and a shared Slack channel. It is a public institution with everyday obligations and unevenly distributed digital maturity.That makes it a useful test case for Microsoft 365 Copilot adoption in the real world. Municipal employees are not all knowledge workers in the same sense. Some spend their days in documents and inboxes. Others operate closer to service delivery, planning, citizen interaction, or field coordination. Their tolerance for AI experimentation will vary, as will the amount of time they can realistically devote to training.
The city’s own framing emphasizes that its strategy elevated data as an enabler of everyday work and decision-making. That language sounds familiar because almost every public-sector digital strategy now says something similar. The harder question is how data becomes useful without turning employees into reluctant data analysts or exposing them to tools they do not trust.
Copilot offers one possible answer: put generative AI into the productivity applications employees already use. But embedded AI can also create the illusion of simplicity. If a tool appears inside Word or Teams, employees may assume it is just another feature, rather than a new way of querying organizational knowledge. Raisio’s training-first approach acknowledges that the interface may be simple while the institutional implications are not.
There is also a cultural dimension. Public administration has good reasons to be cautious. It operates under legal obligations, political scrutiny, budget constraints, and high expectations for fairness and transparency. When a private company experiments with AI and saves a few hours on slide decks, the risk may be manageable. When a city uses AI in the flow of public service, confidence and accountability matter more.
Sogeti’s Role Shows the Partner Economy Behind Copilot
Microsoft sells Copilot as a product, but much of the adoption work is flowing through partners. Raisio chose Sogeti, part of Capgemini, citing both Microsoft technical expertise and a human-centered approach to change. That pairing is revealing. The city did not appear to want a partner that would merely explain features; it wanted one that would help reshape work habits.This is where the Copilot economy becomes less about AI models and more about organizational consulting. The software is standardized, but the deployment is not. A municipality, a law firm, a hospital, and a manufacturer may all buy Microsoft 365 Copilot, yet each has different data risks, workflows, language needs, compliance obligations, and employee anxieties. The partner’s job is to translate a general-purpose assistant into credible local practice.
The Raisio program did that through assessment before instruction. Sogeti examined the needs of different employee groups before designing the learning path. That detail sounds ordinary, but it is precisely what many AI rollouts skip. Too many deployments begin with tool demos and end with a vague hope that users will “find use cases.”
Role-specific workshops are the antidote to that. Employees learn faster when the examples resemble their actual work. A prompt-writing lesson about marketing copy will not help a municipal employee understand how to summarize a council document, draft an internal memo, or extract action items from a Teams meeting. AI training becomes credible when it respects professional context.
The partnership also produced AI usage guidelines for the city. That may prove more durable than any single workshop. Guidelines give employees a shared language for what is acceptable, what requires caution, and where privacy and security boundaries sit. In an environment where unofficial AI tools are only a browser tab away, internal rules can reduce both fear and improvisation.
The Safest AI Rollout Is the One That Admits People Are Nervous
Raisio’s project explicitly recognized that employees would arrive with different skill levels, different expectations, and different degrees of skepticism. That is not a barrier to adoption; it is the terrain on which adoption happens. The worst AI programs treat hesitation as resistance. Better ones treat it as data.Pre- and post-project surveys reportedly showed that training reduced hesitation and helped employees understand both the potential and limitations of AI. That last phrase is important. A successful Copilot rollout does not convince everyone that the tool is magic. It teaches them when the tool is helpful, when it is unreliable, and when human judgment must remain firmly in control.
This is especially important for generative AI because the failure mode is not always obvious. Traditional software usually fails visibly: an error message, a crash, a missing field. AI can fail fluently. It can summarize with misplaced emphasis, draft with unwarranted confidence, or overlook context that a human would catch. Employees who are trained only to be impressed are poorly prepared for that reality.
Raisio’s emphasis on a safe learning environment is therefore more than motivational language. People need permission to experiment, ask basic questions, and compare results without embarrassment. The participant who described starting “from zero” and later using Copilot as an everyday assistant captures the point: adoption is not a binary switch. It is a gradual movement from unfamiliarity to routine use.
That is also why peer learning matters. In workplace AI, the most persuasive training may come from a colleague who has discovered a practical use case. Official training explains the tool; peer examples normalize it. When employees share what worked, what failed, and what saved them time, AI stops being an executive initiative and becomes part of the organization’s practical vocabulary.
The Real Productivity Prize Is Not Drafting Faster
The most obvious Copilot use cases are document production, summarization, search, email management, and communications support. Raisio reported that employees began using Copilot to produce documents more easily, summarize large amounts of information, search more efficiently, and manage communication. Those are credible early wins because they live close to daily friction.But the deeper productivity prize is not simply faster writing. It is reducing the cognitive tax of navigating fragmented information. Public-sector employees often spend time locating the right version of a document, reconstructing decisions from email threads, turning meetings into actions, or converting policy material into usable language. AI can help with that, provided the underlying information is accessible, permissioned, and reasonably well managed.
This is where municipal AI adoption intersects with a much older IT problem: information architecture. Copilot cannot magically fix inconsistent document storage, obsolete files, weak metadata, or overly broad permissions. In some cases, it will expose those weaknesses faster. A tool that can retrieve and synthesize information across Microsoft 365 is only as safe and useful as the environment it is allowed to see.
That reality should temper the enthusiasm around early success stories. Nearly 100 trained employees is meaningful for Raisio, but it is not the same thing as full organizational transformation. The training program creates a foundation. The harder work comes later, when the city must decide which processes deserve deeper automation, which datasets need better governance, and which AI uses are inappropriate for public administration.
Still, early routines matter. If employees learn to use Copilot for low-risk tasks, they build judgment before the organization moves into more consequential workflows. Summarizing meeting notes, drafting first-pass documents, and organizing communications are not trivial tasks, but they are a safer starting point than automated decision support or resident-facing AI services. Raisio’s gradualism is a feature, not a lack of ambition.
Governance Is Where the AI Strategy Becomes Real
Every organization now says it wants responsible AI. The phrase is so overused that it risks becoming decorative. Raisio’s case is useful because it shows what responsible AI looks like in mundane form: training materials, usage guidelines, privacy awareness, security practices, and a measured rollout.Governance in this context is not a committee that says no. It is the set of conditions that lets employees say yes safely. Workers need to know whether they can paste certain information into prompts, when they should verify outputs, how to treat AI-generated text, and whether Copilot’s access reflects existing permissions. They also need clarity on what kinds of tasks remain human-owned.
For a city, that clarity is essential. Municipal government must maintain public trust. If employees use AI to improve internal efficiency, residents may welcome the benefits. If AI appears to influence decisions without transparency, or if sensitive information is mishandled, trust can evaporate quickly. The line between internal productivity and public accountability must be deliberately drawn.
Microsoft’s enterprise pitch for Copilot leans heavily on security and integration with Microsoft 365 controls. That matters, but it does not remove the need for local governance. A technically secure platform can still be used unwisely. A permissions model can still reflect years of accumulated access sprawl. A user can still overtrust a confident summary.
Raisio’s co-created AI guidelines suggest a better model: make policy part of adoption rather than an after-the-fact restriction. Employees are more likely to follow rules they understand in the context of their own work. The point is not to bury AI under bureaucracy. It is to make the boundaries visible enough that experimentation can continue without reckless improvisation.
Copilot Forces IT to Become an Education Function
For years, enterprise IT has tried to move from ticket-taking to strategic enablement. AI may finally force that transition, whether IT departments are ready or not. A Copilot rollout is not primarily a deployment project in the old sense. It is a continuous education program wrapped around identity, data governance, security, and workflow redesign.Raisio’s CIO framed the employer’s role as enabling people to succeed, making work smoother, and supporting well-being. That is a notable shift in emphasis. The value proposition is not only “do more with less,” the phrase that haunts public-sector technology projects. It is also “reduce unnecessary friction so employees can focus on meaningful work.”
That framing is politically and operationally smarter. Workers are understandably wary when AI is introduced as a productivity mandate. If the message is that AI exists to squeeze more output from the same headcount, employees may comply without trust. If the message is that AI can remove routine burdens while preserving professional judgment, the organization has a better chance of honest adoption.
But that promise must be earned. Training cannot be a one-off webinar. It must account for new features, changing policies, emerging use cases, and employee feedback. Microsoft is evolving Copilot rapidly, and the line between assistant, agent, automation layer, and workflow orchestrator will keep shifting. IT and digital teams will need to help employees keep up without turning every update into a new anxiety cycle.
In that sense, Raisio’s open access to materials after the project matters. Self-study resources allow learning to continue beyond scheduled sessions. They also help employees revisit concepts when they encounter a real task, which is often when training finally becomes relevant.
Automation Is the Next Battle, Not the Next Feature
The third part of Raisio’s learning model introduced opportunities for automation and the future of knowledge work. That may sound like a preview module, but it points to the next phase of enterprise AI adoption. Once employees are comfortable using Copilot to draft, summarize, and search, organizations will naturally ask what can be automated end to end.This is where the risk profile changes. A human asking Copilot to summarize a document can review the output before using it. An automated workflow that routes information, drafts responses, updates records, or triggers actions requires stronger controls. The move from assistant to agent raises the stakes.
For municipalities, that transition must be handled carefully. Automating internal administrative steps may deliver real benefits. Automating anything that affects residents, entitlements, inspections, services, or official decisions requires a much higher bar. Even if AI is not making the final decision, its role in shaping the information available to decision-makers deserves scrutiny.
Raisio’s gradual path gives it a better chance of navigating that future. Employees who understand AI’s limits are more likely to design sensible automation. Leaders who have listened to staff concerns are more likely to spot where process change is needed before technology is added. IT teams that have built guidelines early are better positioned to extend them into more complex scenarios.
The danger for every organization is the temptation to treat automation as an inevitable next step. It is not. Some workflows should be accelerated, some should be redesigned, and some should remain deliberately human. The point of AI maturity is not to automate everything. It is to know the difference.
Microsoft Wins When Customers Discover the Hard Part Themselves
The Raisio story is also a Microsoft story, even though Microsoft is not the main actor in the case study. Copilot’s long-term success depends on customers learning that adoption is organizational, not merely technical. That is both an opportunity and a problem for Microsoft.On the opportunity side, Copilot sits in software many organizations already use. That gives Microsoft a distribution advantage few competitors can match. If employees live in Microsoft 365 all day, an AI assistant embedded there has a natural path into everyday work. For public-sector organizations already standardized on Microsoft tools, Copilot may feel like the least disruptive AI option.
On the problem side, that same familiarity can encourage underinvestment in rollout. Because Copilot appears inside known applications, leaders may underestimate the need for training and governance. They may assume employees will simply learn by doing. Some will, but many will not, and the unevenness can create disappointing usage metrics, inconsistent practices, or quiet workarounds using unsanctioned AI tools.
Microsoft therefore needs stories like Raisio’s. They demonstrate that Copilot adoption can work when paired with structured learning, partner support, and executive sponsorship. They also subtly shift responsibility back to the customer: if the tool disappoints, did the organization prepare its data, train its people, and define its policies?
That is fair only up to a point. Vendors should not get to sell transformation and then blame customers for needing transformation work. But the practical lesson remains: Copilot is not self-implementing. The organizations that get value will be the ones that treat adoption as a program, not a toggle.
The Raisio Model Is Modest, Which Is Why It Matters
There is a refreshing lack of grandiosity in the Raisio project. Nearly 100 employees trained is not a global megadeployment. The reported benefits are practical rather than revolutionary. Documents became easier to produce. Large information sets became easier to summarize. Search and communication improved. Employees became less hesitant.That modesty is precisely why the case is worth attention. Most organizations do not need another breathless claim that AI will reinvent work overnight. They need examples of how to begin without breaking trust. Raisio’s project offers a pattern: align AI with strategy, assess employee needs, train in stages, build guidelines, allow self-paced learning, and keep communication open.
The city’s leadership also avoided a common mistake: separating technology from culture. It chose a partner partly because technical expertise and change management were treated as inseparable. That should be obvious by now, but in enterprise IT it still is not. Too many projects divide the world into deployment on one side and adoption on the other, as though users are an afterthought.
The better view is that adoption is the product. A Copilot license sitting unused, misused, or feared is not a productivity tool. It is a recurring cost. A Copilot license in the hands of an employee who understands when to use it, how to verify it, and what not to feed it is something different: a small but compounding change in how knowledge work gets done.
Raisio’s path also has the advantage of being replicable. Not every city has the same budget, partner access, or Microsoft footprint, but the principles travel well. Start with work, not features. Train by role, not slogan. Publish rules early. Encourage peer learning. Measure confidence as well as usage.
Raisio’s Lesson Is That AI Adoption Has to Feel Ordinary Before It Becomes Transformational
The concrete lesson from Raisio is not that every municipality should copy its exact training program. It is that public-sector AI projects need to become ordinary enough for employees to use safely before leaders ask them to become transformational. The city’s approach offers a grounded template for organizations that want AI benefits without pretending that culture, security, and trust will take care of themselves.- Raisio began its Copilot adoption in autumn 2025 as part of a wider strategy to make data a daily asset in municipal work by 2026.
- The city trained nearly 100 employees while keeping learning materials available more broadly across the organization.
- The rollout used a three-stage model that combined introductory training, role-specific workshops, and an early view of automation opportunities.
- Sogeti’s role shows how much of the Microsoft 365 Copilot market depends on partner-led change management rather than software deployment alone.
- The city’s most important decision was to create AI usage guidelines alongside training, making privacy and security part of daily practice rather than a separate compliance lecture.
- The early benefits were practical and believable: better drafting, faster summarization, more efficient search, and improved communication management.
References
- Primary source: Capgemini
Published: 2026-06-24T08:12:08.618155
The City of Raisio launches a people-first path to AI with Microsoft 365 Copilot
The City of Raisio partnered with Sogeti to launch a new strategy, setting in motion a people-first AI journey supported by a safe learning environment and shared learningwww.capgemini.com - Related coverage: sogeti.com
The City of Raisio launches a people-first path to AI with Microsoft 365 Copilot - Sogeti Global
Discover how the City of Raisio adopted AI with Microsoft 365 Copilot through a people-first approach and a structured rollout.www.sogeti.com - Official source: adoption.microsoft.com
Microsoft 365 Copilot – Microsoft Adoption
Hear directly from Microsoft leaders on how to lead Copilot adoption in your organization.adoption.microsoft.com - Related coverage: windowsforum.com
Raisio’s Microsoft 365 Copilot Rollout: Workforce-First AI for Finland City Hall | Windows Forum
The City of Raisio in Finland began a Microsoft 365 Copilot adoption program in autumn 2025 with Sogeti, part of Capgemini, training nearly 100 municipal...windowsforum.com - Related coverage: sogeti.es
Maximiza el uso de M365 Copilot en tu organización - Sogeti ES
El World Quality Report es el estudio de investigación más grande de la industria, que ofrece una evaluación completa del estado actual de las prácticas de ingeniería de calidad a nivel global y en diversas industrias.www.sogeti.es - Related coverage: sogeti.us
Microsoft - Sogeti US
With 20+ years of partnership, discover how Sogeti and Microsoft collaborate to deliver cutting-edge cloud, data, and AI solutions.www.sogeti.us
- Official source: microsoft.com
Microsoft 365 Copilot for executives: Sharing our deployment and adoption journey at Microsoft - Inside Track Blog
Read our readiness guide that walks executives through how we deployed and are driving adoption of Microsoft 365 Copilot internally at Microsoft.www.microsoft.com - Related coverage: academy.sogeti.nl
Microsoft Office 365 Copilot - Sogeti Academy
academy.sogeti.nl
- Official source: news.microsoft.com
- Related coverage: ajaia.ai
Copilot Cowork Training | Microsoft 365 Frontier Agents | Ajaia
Prepare teams to use Microsoft 365 Copilot Cowork for delegated work, checkpoints, approvals, Microsoft 365 actions, governance, and review.ajaia.ai
