Providing reliable citizen services requires huge administrative effort at every level of government. Citizens want more flexibility, accessibility and visibility from public services, while the departments delivering them face constant pressure to do more with less. Many organisations are turning to AI to help – Gartner predicts at least 80 per cent of governments will be using AI agents to automate routine decision-making by 2028. But disparate AI projects tend not to deliver the desired results.
“Many governments don’t struggle to start AI projects; the challenge is turning early wins into something the institution can run on,” says John Doyle, senior director of worldwide government, defence and intelligence at Microsoft.
“Typically, one department launches a pilot, another picks different tooling, a third sets up its own governance, and in many cases projects can become isolated. That can create enterprise-level complexity with only departmental-level impact, leaving citizens with multiple front doors to get one answer.” And the workforce sees AI as something IT is experimenting with, not something the organisation runs on.”
The City of Burlington in Canada has rebuilt its permitting experience around residents instead of departments (Photo: Adobe Stock/salil)
In addition, many governments tend to focus on AI as a tool to rapidly surface risks, recommendations and operational signals. But when those insights are left languishing in dashboards or disconnected systems, they can’t drive effective actions. Instead, they emphasise another challenge lurking in the cracks – decision latency.
“Insight alone doesn’t improve outcomes – decisions do,” says Doyle. “Decision latency occurs when there is a gap between knowing and acting. AI didn’t create it. AI exposed it and it shows up everywhere: emergency response, cybersecurity, permitting, benefits administration and infrastructure resilience. Often the problem isn’t whether someone had the right information. It’s whether the institution could move on it fast enough. Delays between insight and action erode trust, especially when people see the institution knows something but isn’t acting.”
There is an alternative. When governments take an organisation-wide view of AI strategy, they have the potential to transform citizen experiences, empower workers and support reliable and compliant operations at scale.
“Governments that are advancing in this area often treat AI less like a portfolio of experiments and more like shared infrastructure: shared identity, governance and security and a consistent way for data and decisions to move across the institution,” says Doyle. “They aren’t designing around tools, but around decisions. How quickly can we respond? How consistently can we act? How do we get better over time? That’s the shift from experimental to operational, and from pilot to platform.”
This connected thinking supports decision intelligence.
“Decision intelligence is about closing the gap between signal and action,” explains Doyle. “It connects an AI-generated insight to an actual decision, gives it an owner and creates a feedback loop to capture what happened next. That loop matters in government because decisions carry accountability. Citizens need transparency, regulators need auditability and institutions need consistency across departments and services.”
What does this approach look like on the ground? For citizens, decision intelligence allows governments to organise services and experiences around the people who use them, instead of agency boundaries.
“Most public services still ask people to understand how the institution is internally structured: departments, workflows, portals, approvals and disconnected systems,” says Doyle. “Digitising those experiences helps, but it doesn’t eliminate the friction; it just moves it online.
“AI and decision intelligence bring the opportunity to redesign the journey itself. Instead of asking citizens to negotiate the organisational chart, governments can engage citizens more proactively, in one conversation, with more relevant and timely responses – or a clear, accountable path to an answer.”
A new solution based on Microsoft Dynamics 365 means that citizens of New Zealand have access to a single online portal where they can apply for a passport as individuals or groups
Governments around the world are already putting this citizen-centric approach into practice. In Canada, for example, the City of Burlington has rebuilt parts of its permitting experience around residents instead of departments, so a citizen can finish what they started without chasing approvals across silos. New Zealand’s Department of Internal Affairs has created a single, trusted route into services that used to be scattered across multiple agencies. Meanwhile, Buenos Aires uses generative AI to meet residents in the language and channels they use.
“These efforts aren’t just automating steps; they're redesigning how the process feels from the citizen’s side,” says Doyle. “Citizens don’t care how government is internally organised. They care whether the experience feels coherent, accessible and human. When governments build on a trusted, sovereign-ready foundation, they create a new interface between people and their government.”
The approach can also empower the people delivering those services, leaving them free to focus on their most valued work: helping people. In a sector beset by budget cuts, labour shortages and increased workloads, Doyle says outdated workflows are the real problem.
“Significant value can come when AI reduces friction in the work people already do every day,” says Doyle. “For frontline workers, that means surfacing the next-best action or reducing administrative burden during citizen interactions. For back-office teams, it means simplifying documentation, summarisation, reporting and coordination work. For leaders, it means moving from fragmented information to operational clarity.
“Crucially, if AI becomes one more system for employees to stop and learn, you've made the problem worse, not better. Augmentation is most effective when it integrates seamlessly into the flow of work – built with people, not dropped on them. It should give civil servants their craft back.”
In Buenos Aires, Argentina, the Boti app, built using Microsoft Azure OpenAI Services, manages approximately two million natural language queries per month (Photo: iStock/Jeremy Poland)
Getting AI right for employees provides a foundation for every other area of transformation. In fact, employee sentiment has become as important as traditional productivity measures for government AI deployments.
“It is difficult to redesign citizen experience, operate reliably at scale or build institutional trust if the workforce does not trust the tools they use every day,” says Doyle. “Workforce adoption is where decision intelligence either takes root or stalls.
“In one reported deployment, the UK Government Digital Service rolled Microsoft 365 Copilot out to approximately 20,000 civil servants, and a majority of respondents indicated they wouldn’t want to return to working without it. That isn’t really a productivity statistic. It’s an engagement measure, and engagement is the harder one to earn in any workforce, especially in public service. We’re seeing the same pattern across local government, including organisations like Buckinghamshire and Aberdeen councils, where the focus has been on reducing administrative burden and giving employees more time for higher-value work.
“In these examples, AI was implemented with the workforce and the deployments were treated as workforce enablement, not rollouts. Adoption can indicate the technology is delivering value. Sustained use suggests the workforce trusts it. That’s the difference between a project and a capability and it’s what makes the improvement compound over time instead of fading after the initial rollout.”
A Copilot trial of 20,000 civil servants in the UK showed they could on average save nearly two weeks of time per year (Photo: iStock/Leo Patrizl)
Ultimately, trust in government is hard-won and easily lost – and administration systems must continually inspire confidence among users and regulators.
“Governments operate against some of the highest trust requirements in the world,” says Doyle. “Security, privacy, compliance and sovereignty must be foundational to any AI strategy, not added later.”
Microsoft provides the platform layer through its sovereign-ready cloud, integrated security, governed AI and robust compliance stance. And its partners bring the expertise to build tailored solutions addressing governments’ unique needs.
“Governments don’t modernise through platforms alone,” says Doyle. “Our partner ecosystem helps translate AI capabilities into differentiated value and lasting public outcomes. Microsoft partners bring agency knowledge, regulator fluency and deep vertical expertise in areas like citizen services, tax, benefits, justice, public finance and defence and intelligence. They bring significant experience and understanding of local policy environments and operational workflows.
“Governments leading on AI build institutional systems citizens and regulators trust – and they do it through partners. Recent examples include Esri’s ArcGIS on Azure, which gives public safety agencies spatial intelligence – predicting traffic, optimizing emergency routes and monitoring infrastructure in real time. Accenture’s AI Refinery runs AI agents across the citizen service-delivery lifecycle, from intake to resolution. And BlueVoyant delivers cross-estate managed detection and response on Microsoft Sentinel and Defender, operating inside the customer’s own tenant.
“In each case, the result is operational systems that are sovereign by design, productive by outcome and secure by default.”
Partner perspectives
We asked Microsoft partners how they are using Microsoft’s cloud and AI technology to help governmental organisations improve operational processes, while delivering better employee and citizen experiences.
“Accela technology is developed on Microsoft’s cloud to help government agencies modernise permitting and licensing in ways that are practical, transparent and accountable,” said Noam Reininger, CEO of Accela. “By embedding AI directly into core administrative workflows, we help agencies automate document intake, surface issues earlier in the review process and provide real-time visibility for both staff and applicants. Importantly, AI augments – rather than replaces – human decision-making, enabling agencies to manage capacity constraints, reduce backlogs and improve service experiences while maintaining trust, compliance and oversight. The result is faster, more predictable processes that improve outcomes for employees, residents and businesses alike.”
Noam Reininger, CEO, Accela
“Coretek modernises government operations using Microsoft AI Foundry, Azure Government Cloud and intelligent process automation,” said Brian Barnes, chief technology officer at Coretek. “We deployed AI-powered document processing, citizen-facing virtual agents and Microsoft Copilot Studio workflows that eliminate paper-based bottlenecks and dramatically improve service delivery speed. A large municipal government in Michigan, USA, reduced permit processing time by 50 per cent and resolved 38 per cent of citizen enquiries autonomously through an AI-driven self-service portal. Agency employees now redirect effort from administrative tasks to complex, high-impact citizen services. The outcome: faster government, better citizen experiences and empowered employees – all built on a compliant, secure Microsoft AI foundation.”
Brian Barnes, Chief Technology Officer, Coretek
Discover more in the Summer 2026 Issue of Technology Record. For more content like this, subscribe to the print edition or free digital edition of the quarterly Technology Record magazine.