Singapore’s enterprises have rebuilt their AI foundations and now sit above the global benchmark. The 2026 Enterprise AI Maturity Index reveals that the nation’s AI maturity score has rebounded strongly to 53 out of 100, surpassing the global average of 51 and recovering from a low of 34 in 2025.
Agentic AI adoption has more than doubled, rising from 22% in 2025 to 51% in 2026. Yet beneath these encouraging headlines lies a critical paradox: while most enterprises are using AI to help individuals work faster, only 10% have redesigned end-to-end workflows.
This gap between adoption and execution—what ServiceNow calls the “chaos gap”—represents the defining challenge for CIOs across Southeast Asia.
In an exclusive interview with FutureCIO, Colin Tan, ServiceNow’s country manager in Singapore, explains why closing this gap requires more than technology investment. It demands a fundamental rethink of how work flows through the enterprise.
Building on imperfect foundations
One of the most persistent barriers to AI transformation is data quality. According to the Singapore country sheet, 68% of Singapore executives cite inadequate data accuracy and access as their top AI challenge, while 58% cite data privacy and security concerns—above the global average of 49%.
Tan argues that organisations must abandon the myth of perfect data. “AI is only as effective as the data and context behind it,” he says. “The priority is really connecting trusted data to the workflow where AI needs to act and not waiting for the data to be perfect.”
He references ServiceNow’s chief product officer, Amit Zavery, who recently visited Singapore and offered a blunt assessment:
“There’s no single company that has been able to reset anything to pure underlying data structure and keep up with that. We’re not going to wait till a day where everything is cleaned up before we can do AI.” Amit Zavery
Tan describes fragmented data and system sprawl as “the permanent condition” of modern enterprises. The more important question, he suggests, is not when data will be ready, but “how do we build a workflow that can operate on top of this imperfect and distributed data while we continue to maintain the governance and visibility when we use the AI?”
He points to Griffith University in Australia as proof that progress is possible without perfect data. By consolidating fragmented systems into a single AI platform and connecting AI to data and workflow, the university achieved an 87% increase in self-service, a 43% improvement in first-contact resolution, and a 31% reduction in call volumes.
Infrastructure: Building roads for tomorrow’s traffic
As agentic AI proliferates, the demands on enterprise infrastructure intensify. Tan warns that AI is “putting a lot of stress on the current enterprise infrastructure,” citing “a huge cost spike in memory, GPUs, servers, and so on.”
He poses a fundamental question for CIOs: “Are we building roads for the traffic we have today, or the traffic we know is coming?”
Drawing on a recent speech by Singapore’s Minister Josephine Teo, Tan uses Changi Airport’s Terminal 5 as a metaphor. “A new terminal alone doesn’t do the job. It’s very much the hardware and the software have to move together.”
Singapore’s AI maturity score recovered—from 45 in 2024 to 34 in 2025, then up to 53 in 2026—showing enterprises have strengthened their foundations. However, Tan cautions that “infrastructure alone is not enough.”
“The real opportunity for us is to make sure that every step or every layer of your AI, whether you’re looking at infrastructure, data, workflow, governance, as well as security, works together, rather than creating another layer of fragmentation.” Colin Tan
The execution gap from automation to orchestration
The most significant finding in the 2026 Index is the disparity between agentic AI adoption (51%) and end-to-end workflow redesign (10%). Tan calls this “arguably the biggest execution gap that we see in the Singapore AI journey.”
The stakes are high. According to the Singapore country sheet, organisations that redesigned their workflows around AI saw productivity gains about four times higher than those that layered AI onto existing processes.
Tan highlights Standard Chartered Bank as a pacesetter that has embraced this principle. Rather than adding AI to individual steps, the bank reimagined its employee onboarding journey end to end. The transformation is expected to reduce onboarding effort for hiring managers by 35% while delivering approximately 20% productivity gains across HR teams.
“The objective is very much to identify, for any organisation today, a few of the high-value workflows across their enterprises,” Tan explains. “Redesign them end-to-end, improve the business impact, and then scale from there.”
The threat of agent sprawl
As agentic AI adoption accelerates, a new risk emerges: agent sprawl. Tan describes this as “the uncontrolled spread of AI tools and models and even agents across the enterprise,” sometimes including “shadow IT assets where you’re not aware of.”
He offers a vivid analogy: “You can have hundreds of AI agents running across the enterprise. But without orchestration, you risk getting into AI sprawl. It’s like having an orchestra with no conductor. Everyone may be playing, but you don’t necessarily get good music.”
Citing a Ministry of Manpower study, Tan notes that among larger firms that have adopted AI, 56.1% cite integration complexity and 55.4% cite data security as barriers. “The answer is therefore not to deploy more disconnected AI tools, but to create greater visibility, integration and control over how AI operates.”
Security concerns compound the challenge. Tan reveals that “an average enterprise runs probably more than 70 security tools, and the machine identity is doubling, roughly double every 18 months. So the attack surface is exploding.”
He emphasises the need to manage “the humans, as well as non-human identities, with the right access controls put in place” and the ability “to cut off the AI agent when it goes rogue.”
Governance: The prerequisite for autonomous AI
Governance has become a board-level concern. Yet Tan observes “a significant gap between the governance intent and the operational execution.”
He urges organisations to view governance as an enabler, not a barrier: “Strengthening AI governance and putting in the guardrails as enablers is not a barrier to AI innovation.”
Critically, he stresses that “autonomous AI should not mean AI on autopilot.” Organisations need “a structured AI control tower“—a centralised platform to discover, secure, govern, and measure the value of AI assets.
Singapore has made notable progress. According to the country sheet, 28% of Singapore enterprises have formal AI testing, auditing, and risk processes in place, compared with 20% globally. The IMDA also became the first regulator globally to publish a governance framework for autonomous AI systems in January 2026.
Tan praises Singapore’s leadership but notes that “the majority have yet to formalise some of these processes.” He adds that enterprises are “keeping pace with the IMDA framework by adopting a phased rollout” and implementing “human-in-the-loop checkpoints” and “multi-layer testing.”
Measuring what matters: Beyond cost savings
As AI shifts from efficiency to revenue creation, traditional metrics become inadequate. Tan acknowledges that “Singapore AI investment has indeed grown about 108% year-on-year. But investment alone does not tell us whether an organisation is becoming more productive or transforming how it operates.”
He points to the four-fold productivity advantage of workflow redesign and suggests a broader set of metrics: “Productivity gains across teams and workflows. Reduced manual effort, or even process friction. Faster resolution and service delivery. And improving the customer’s experience, end-to-end.”
Standard Chartered’s tangible targets—a 35% reduction in hiring manager onboarding effort and approximately a 20% productivity gain across HR teams—illustrate how pacesetters translate AI investment into measurable business outcomes.
Learning from the pacesetters
The 2026 Index identifies a cohort of “Pacesetters”—approximately 21% of organisations with an AI maturity score above 60. What distinguishes them is not budget or ambition, but foundation.
“The pacesetters are not asking where they can put AI,” Tan says. “They’re asking what work should look like if AI is part of the team.”
He outlines five strategies that set pacesetters apart:
- Set a shared strategic vision beyond efficiency. Frame AI as business reinvention, not just a productivity tool.
- Control your data to unleash your AI. Don’t wait for perfect data; build the connective tissue.
- Build the workforce your AI strategy demands. Invest continuously in human capabilities alongside AI capabilities.
- Accelerate governance and close the gap. Embed trust and transparency into AI processes.
- Move beyond automation to orchestration. Redesign workflows end to end rather than automating isolated tasks.
2027’s path forward
For CIOs across Southeast Asia, the message is clear: escaping the chaos gap requires discipline, governance, and a willingness to redesign work rather than digitise it.
Tan’s final advice is pragmatic: “There are certain things where, if there’s a deterministic need, it doesn’t necessarily come from AI. A simple workflow will be able to get certain tasks done.”
The key is clarity of purpose.
“Be very clear in terms of the objective and outcome, and put in place a governance framework that allows you to scale, test out your AI, your use cases, and once you test it out, then that’s where you scale.” Colin Tan
Singapore’s recovery in AI maturity demonstrates that foundations can be rebuilt. But as Tan makes clear, the real transformation—the kind that delivers 160% ROI and 6.5x more revenue channels—requires something more: the courage to redesign how work flows through the enterprise. Only then can organisations escape the chaos gap and achieve genuine AI-powered business reinvention.
Click the PodChats player to hear Tan elaborate on what the survey means for CIOs and technology leaders as they steer their organisation through their AI-powered transformation journey.
- Strategy & Ambition: In the ServiceNow study, 68% of organisations cited inadequate data accuracy, access, and management as ongoing challenges. How do we modernise our data architecture to provide the clean, connected, and real-time intelligence AI agents need to act reliably at scale?
- Infrastructure & Scale: As agentic AI generates exponentially more network traffic and computational demand, is our current infrastructure resilient enough to handle this surge, or are we risking a “digital congestion collapse” that undermines performance and agility?
- Workflow Orchestration: While 51% of Singapore enterprises now use agentic AI, only 10% have redesigned end-to-end workflows. How do we shift from automating isolated tasks to orchestrating autonomous, cross-functional workflows that deliver measurable business transformation?
- Taming “Agent Sprawl”: As autonomous agents proliferate across departments, they risk creating new technical debt and fragmentation. How do we architect a unified control plane to gain visibility, manage permissions, and prevent chaotic, disconnected agent deployments?
- Governance & Risk: Governance is the prerequisite for autonomous AI. How do we build a living governance framework that provides continuous oversight, embeds trust and transparency, and ensures accountability before agents operate at scale?
- Singapore’s IMDA published the world’s first governance framework for agentic AI. How are Singapore enterprises keeping pace with the standard the government has set?
- Measuring New Value: As AI shifts from efficiency to revenue creation and new business models, how should we evolve our metrics to accurately capture the value of AI-enabled agility, innovation, and competitive differentiation—not just cost savings?
Lesson from Pacesetters: The ServiceNow study shows that a “Pacesetter” minority that redesigns work around AI can achieve significantly higher ROI than those that layer AI onto existing processes. What can we learn from Pacesetters?
