Artificial intelligence (AI) has become a central theme in digital government strategy. But for many public sector organizations, the challenge is no longer whether to adopt AI – it’s how to move beyond pilot projects to real operational change.
At the ServiceNow AI Summit in London, the company made its case for how that shift will happen. The answer, it argues, is not simply more powerful models or smarter chatbots, but embedding AI directly into the workflows that run organizations.
“AI without workflow is just an expensive advice tool. AI inside workflow can autonomously drive outcomes across government,” Aaron Neil, VP public sector UK at ServiceNow, noted during the event.
Over the past year, ServiceNow has accelerated its push both into generative AI and the public sector – in the UK alone, the sector makes up around a third of the firm’s business.
Within the ServiceNow portfolio, new platforms such as Autonomous Workforce and EmployeeWorks aim to embed AI agents across enterprise workflows, while the company’s acquisition of conversational AI firm Moveworks has expanded its ability to translate natural language requests into automated tasks across systems.
“Public sector customers should be able to get answers and take action as fast as any worker anywhere,” Mike Hurt, group VP for public sector at ServiceNow, noted recently.
The company’s vision is that conversational interfaces combined with workflow automation will allow employees to resolve requests instantly, removing the friction that often slows government processes.
In theory, the technology could transform how large institutions operate. In practice, however, many organizations are still figuring out where AI fits into their operations.
Governance Before Scale
For the public sector, the first phase of AI adoption has often focused less on technology and more on governance. Lee Massie, head of IT at Oxford University Hospitals NHS Foundation Trust, said that his organization deliberately prioritized policy and oversight before expanding AI deployments.
“We started our AI journey a little earlier than many organizations. But it became quite evident quite quickly that we needed structural governance in place to support adoption,” he said.
Healthcare organizations face particular challenges when introducing AI tools, especially when patient interactions and clinical documentation are involved. One of the Trust’s pilot projects involves voice technology that listens to conversations between clinicians and patients and automatically generates structured notes.
The tech has already demonstrated potential productivity benefits. In some clinics, Massie said, clinicians can spend more time with patients because admin tasks are reduced. But deploying such systems safely requires careful implementation.
“There are real concerns around the legitimacy of the transcription in complex medical language,” he explained, noting that clinicians must be confident that automated documentation accurately reflects conversations.
Patients themselves have been largely supportive. During pilot deployments, Massie said, the Trust asked patients whether they were comfortable with voice technology being present in consultations.
“Only one patient said they were uncomfortable,” he said.
The testimony is particularly interesting in light of recent research that shows 51% of the UK worry AI could reduce human contact in public services.
The Orchestration Challenge
As organizations experiment with different AI tools, another challenge is beginning to emerge: how to manage multiple AI systems operating across departments.
ServiceNow believes orchestration platforms will eventually play an important role.
The company’s AI Control Tower concept is designed to monitor AI agents, coordinate their activities, and ensure they operate within governance frameworks. But for many public sector organizations, that stage of maturity is still some distance away.
Richard Michel, chief information and digital officer at the University of London, believes the concept of orchestrating AI systems across an organization is becoming increasingly relevant as vendors embed AI capabilities into their products.
“Everyone is embedding AI into their own products,” he said, adding that the idea of managing how those systems interact is appealing. “The idea of orchestration really excites me. But we’re not quite there yet.”
For technology vendors, this gap between vision and current reality represents both a challenge and an opportunity. While orchestration platforms may represent the next phase of enterprise AI, many organizations are still working through the earlier stages of adoption.
AI at the Scale of National Infrastructure
In some sectors, however, the scale and complexity of operations are already pushing organizations toward AI-driven automation.
Speaking at the summit, Jody Elliott, head of IT risk and sustainability at National Grid, described how AI is beginning to reshape risk management across the organization’s technology estate.
Large infrastructure operators run hundreds of projects simultaneously, each generating changes to systems, code, and operational processes.
Maintaining visibility across all of that activity is extremely difficult for risk teams.
“In large organizations you’ve got multiple agile projects running,” said Elliott. “From a risk perspective, how do I have sight of every story and feature in every backlog? I can’t embed risk people into those discussions because it’s not practical.”
Generative AI is beginning to change that dynamic. By analyzing large volumes of structured and unstructured operational data, AI systems can identify the issues that require attention from human specialists.
“Generative AI gives us the opportunity to analyse all that unstructured data, and turn out the top sort of ten highest-risk things every few weeks,” said Elliott.
That allows risk professionals to focus on the most significant security or regulatory issues rather than attempting to review everything manually.
AI is also helping the company respond to cybersecurity threats. National Grid recently developed an internal AI agent capable of correlating endpoint vulnerability data with information about known exploits and business-critical systems. The system can combine those datasets and prioritize which vulnerabilities require immediate attention.
Elliott said the prototype agent was built quickly and can deliver results in around 90 seconds. Human teams then review the output to verify the analysis.
The real advantage, he explained, lies in combining technical information with business context.
“If you overlay that with HR data, you can identify which devices belong to executives or critical roles and prioritise recovery accordingly,” he said.
Training, Trust, and Risk
Despite the benefits, Elliott emphasised that AI adoption also introduces new risks. One challenge is ensuring employees do not place too much trust in AI-generated outputs.
“There’s a risk that people become subject matter experts when they’re not subject matter experts,” he said.
For that reason, National Grid has introduced AI training programmes across the organization, from senior executives to technical specialists. The aim is to ensure employees understand both what AI can do and where its limitations lie.
Training, Elliott said, cannot be treated as a one-off exercise. As tools evolve, organizations must continually reinforce how they should be used.
Supply chain risk is another growing concern. Many technology vendors are embedding AI capabilities into their products, meaning organizations must assess how those systems operate within their own environments.
A Transatlantic Push
ServiceNow’s ambitions in public sector AI are not limited to Europe.
In the United States, the company is expanding its government presence through its Moveworks platform, which recently achieved FedRAMP Moderate authorization – a certification required for cloud technologies used by US federal agencies.
The approval allows federal departments, defense contractors, and other public sector organizations to deploy the AI assistant platform across their workforce, enabling employees to search and act across enterprise systems through a conversational interface.
ServiceNow says the milestone will allow agencies to automate routine tasks across IT, HR, and procurement while maintaining strict security and compliance standards.
“For federal agencies, security is the bedrock of every mission,” said Damián Hasse, chief information security officer for Moveworks from ServiceNow, adding that the authorization provides a secure path to modernize workflows without compromising security requirements.
The development underlines the company’s broader strategy of positioning its AI platform as infrastructure for government digital transformation.
Final Thoughts
If there was one point of consensus among speakers at the summit, it was that AI will eventually become embedded in almost every organizational process. But the path to that future will likely be gradual. Many institutions are still developing governance frameworks, training programmes, and pilot projects.
For Elliott, the most important step is simply engaging with the technology.
“Just have a go,” he said.
AI, he argued, is already beginning to reshape organizational processes, whether institutions realize it or not. The organizations that benefit most will be those that learn how to guide that transformation – rather than waiting for it to arrive fully formed.