The AI imperative: Revolutionising service management for a smarter future

As Singapore races toward its Smart Nation 2025 vision, its digital ecosystem is undergoing rapid transformation. Chief Information Officers face an existential challenge: adapt rapidly to the integration of Artificial Intelligence and tightening regulations or risk obsolescence.

Recent data reveals that 65% of IT leaders are struggling with significant AI governance skill gaps amid aggressive sustainability targets and escalating cyber threats. For Singapore’s CIOs, 2025 is accelerating them in an increasingly uncertain landscape.

This perfect storm is catalysing a fundamental reimagining of IT Service Management (ITSM). Once relegated to back-office operations, ITSM now occupies the strategic frontline, tasked with building operational resilience and driving sustainable growth.

A recent FutureCIO roundtable, “Accelerate Service Management with AI,” organised in partnership with Freshworks, AWS, and TruVisor, examined this transformation, exploring how AI might forge a hyper-efficient ITSM that serves as a competitive differentiator.

The discussion brought together IT leaders from various sectors, navigating the complexities of AI adoption. Their collective insights reveal AI’s current impact, its future potential, and the substantial hurdles that lie ahead.

AI augmenting human agents: The new frontline

A delegate noted how AI technologies, such as Optical Character Recognition (OCR), have long been allies. “OCR is an AI technology,” he explained, describing its use in processing national exam scripts. “This is just basically to help make things faster, reduce the burden for the teachers.”

He also mentioned nascent research into “AI auto-marking essays,” highlighting the technology’s evolving capabilities while acknowledging the critical element of trust.

A delegate in the hospitality sector acknowledged that the industry’s focus has shifted from customer-facing applications to enhancing internal employee experience. “We are looking at how AI is going to handle some of these questions that are a little less complex… that frees up our agent to do, to take care of people who need help.”

One delegate revealed their organisation’s approach to be deliberate but cautious: “We have deployed AI internally to help with our agents, but we haven’t deployed it to the customer to use. We want our agents to use it to make their work more efficient.”

Their system involves AI drafting email responses with a “human in the loop” who edits and maintains accountability. “You need to train them, just like you need to train AI for them to get good at this,” he added, comparing current AI capabilities to a “recent college graduate.”

Another delegate described how AI is improving internal communications for their 3,000 teachers. “People started using it to draft their emails. I think we can see a drastic improvement in their languages, torn down language.”

He noted that their organisation is exploring how AI can “analyse what they are communicating, where they are lacking, and how we can improve,” particularly in standardising communication quality between teachers and parents.

Bobby Chan, strategic account manager, Asia at Freshworks, highlighted the overwhelming volume of tickets and requests facing service desks. He recounted meeting an IT manager operating on a “WhatsApp basis,” emphasising the unsustainable pressure on service teams. The question becomes: “To what extent can we lighten his workload” with AI?

One delegate distinguished between rule-based automation and genuine AI. “Onboarding system… we call this automation.” He explained that true AI emerges when analysing server logs to “come out with a solution and where to look at an incident ticket” based on models trained with existing knowledge bases.

Enhancing employee experience and operational efficiency

AI-driven internal support shows significant potential for improving employee satisfaction and operational efficiency. Steven Zhang, director of technology at a local bank, shared their experience deploying AI agents for developers.

“In the beginning, they were very happy. First of all, they can drop emails and write some of the code enhancements.” However, the realisation that increased efficiency might reduce manpower needs soon followed.

This sparked crucial debate, with one delegate raising concerns about the future talent pipeline: “If AI does all these things, how should we train the future experts and future senior people… fast forward 30, 50 years in the future, we will be relying so much on AI… what’s going to happen?” This fear of deskilling resonated throughout the discussion.

Felix Chang, a senior head of technology at a global financial institution, acknowledged the excitement surrounding AI, alongside the inherent challenges it presents in regulated environments. “It’s a big challenge to use and deploy AI in the environment,” he stated, citing MAS regulatory compliance as a primary hurdle. “How can we ensure we share with the regulator auditors that whatever we do with AI does not have any impact on our business or customer?”

Despite this, Chang sees clear benefits in areas like analysing false alerts to “improve the efficiency, to really identify what is brought along, what is the real problem getting to tackle.”

Keith Liew, regional account director at Freshworks, addressed data security concerns: “That is just something that you have to go through for InfoSec to understand what has been injected… As long as you get across that hurdle, then the next question is, how do we build trust with the people using the AI?”

He shared an anecdote about an operations manager whose team member spent two hours every Monday producing SLA reports — a task ripe for AI automation.

Consistency, complexity, and the proactive AI

A key aspiration is for AI tools to deliver consistent employee support during high-demand periods and assist with complex queries.

Victor Tan, IT director at Deluge Fire & Protection, emphasised precision requirements: “As an engineering company, if you adopt AI solutions in the engineering world, it has to be very precise and accurate.”

He outlined challenges in preparing decades of legacy data for AI, though they’re experimenting with AI note-takers for project meetings: “Instead of having someone to record the minutes… they have a note taker recording it… so that at the end of the day there’s a trace  and record which can be utilised and allow the project team to perform more productive reporting.”

A delegate at the roundtable confirmed that they are “already adopting it in everything… in several phases in our company.” The organisation utilises third-party tools with embedded AI capabilities. “In terms of interpreting tickets and resolving tickets, I think it helps us in turning around resolution,” though the delegate acknowledged “some false positives in terms of AI as well, and some limitations,” particularly with unstructured data like images.

Freshworks’s Chan identified key challenges in adopting agentic AI: ensuring connectivity to different datasets “to provide the AI with sufficient context” and the “significant cost involved in just experimenting with it.” He noted difficulties in “convincing business stakeholders AI is the future” when highlighting the experimentation costs.

Trust, accountability, and the human element

Trusting AI-driven recommendations for complex issues remains a significant hurdle.

A delegate expressed scepticism about full automation: “Definitely? So that’s no way. At this point.” He advocated for a “hybrid” approach, where AI acts as a “buddy” or advisor. “Assume there’s something a person sitting next to you and advising you, but don’t be the person to answer.”

Stanley Aw, head of IT for a large transportation business, perceived AI as “a tool… like a knife. If you know how to build it correctly, you get to cut things properly. But if you don’t know how to build it correctly, you probably got someone else.” He cautioned against blindly following AI advice, “definitely not for now.”

Freshworks’s Liew suggested that for internal infrastructure and norms, AI recommendations could surpass human judgment, given AI’s ability to process vast data and identify system relationships. “The AI will be able to pick up anomaly… You will probably show what the root cause is, and then you can show where the sources are.”

A critical point raised by multiple participants was the issue of accountability. “Who’s accountable — this is the biggest challenge,” commented a delegate. The “human in the loop” approach — where AI assists, but humans make final decisions and bear responsibility — appears most viable for now.

Conclusion: AI as co-pilot, not autopilot

The roundtable revealed a service management landscape on the precipice of profound AI-driven transformation. What emerges isn’t a vision of AI rendering human agents obsolete but rather one of AI as a sophisticated co-pilot — handling mundane tasks, analysing complexity, predicting probabilities, and personalising experiences.

Yet the human remains essential for navigating ambiguity, exercising ethical judgment, building genuine trust, and, crucially, accepting accountability.

The challenge for CIOs isn’t simply deploying AI but integrating it into the organisational fabric in ways that enhance human capability rather than diminish it. This requires fostering continuous learning cultures, addressing concerns about job displacement, and navigating the complex issues of data privacy and algorithmic bias.

Successful organisations will view AI not as a silver bullet but as a powerful, versatile tool that, wielded wisely, can propel service management into a new era of intelligence and efficiency.

As one delegate aptly noted, AI’s progress is “crazy” fast; today’s conclusions may become tomorrow’s outdated assumptions.

The only certainty is that the AI imperative in service management has evolved from a question of ‘if’ to an urgent ‘how fast’ and ‘how smart’. The future belongs to those who harness this transformative power without losing sight of the human core it ultimately serves.