From Pilots to System-wide Transformation: SG Government’s Next Leap in AI

In policymaking, firsts are important, but they are only the beginning. In 2025, the Singapore government became the first in Asia to deploy agentic AI for public services, enabling public servants to “focus on higher-value work” as highlighted by Prime Minister Lawrence Wong at the G20 Summit. This pioneering spirit also means Singapore is actively navigating the urgent policy questions around accountability, transparency, and human oversight posed by AI agents, requiring forward-looking frameworks to guide their safe and effective integration.

Building on this foundation, 2026 will mark Singapore's shift from AI curiosity to organizational commitment in public services, advancing further towards its National AI Strategy. The potential prize – a leap in productivity and public responsiveness – is within reach, but claiming it will require deeper structural reforms.

Breaking down silos: Strategic cross-agency coordination

A major barrier to scale is the fragmented state of AI adoption across agencies. While pockets of innovation exist, according to a whitepaper by IDC and Dell, titled “Building a Leading Sovereign AI Nation,” 47% of Singapore public agencies surveyed cited lack of strategic plans and clear business cases for Sovereign AI adoption as one of the key factors holding back Sovereign AI ambitions. In an era in which AI sovereignty is becoming a national security priority worldwide, this gap must be closed.

To scale up AI capability, the public sector will require a unified, cross-agency execution strategy to synchronise investments and accelerate learning. GovTech Singapore’s agentic AI architecture and responsible system aim to design playbooks that point agencies in the right direction. This approach is essential for preventing excessive time and resources expenditure resulting from parallel development. The government can also minimize overlapping efforts and incentivize the sharing of solutions, monitoring progress through sequenced milestones.

Crucially, rigor must precede rollout: agencies must define robust success metrics before pilots begin. By quantifying expected benefits like time saved, cost avoided, risks reduced, and identifying scale potential at the outset, agencies can demonstrate tangible value. When agencies can confidently prove what works, national progress accelerates.

Modernizing IT systems as a strategic enabler

Scaling AI requires an infrastructure that is fit for purpose. Legacy systems remain a stubborn obstacle. Research has shown that 61% of public-sector leaders cited inadequate data infrastructure as a barrier to innovation. This is hardly surprising. Advanced AI systems require secure, interoperable, and scalable data pipelines and cannot be layered on top of outdated technology.

Modernization must therefore begin with streamlining procurement processes so that agencies can move rapidly from identifying solutions to deploying them. Bureaucratic postponements can be costly, as they delay benefits and let technology stagnate. To maximize return on investments, governments should prioritize developing shared infrastructure, from common APIs to national data commons and reusable foundational models.

Public–private partnerships will be critical in facilitating this transition. By working with industries to adopt ready-to-use platforms, the government can leapfrog years of incremental upgrades, ensuring that Singapore’s public sector stays at the cutting edge. This deeper collaboration, poised to expand in 2026, will align national objectives with industry expertise and position Singapore as a global leader in exporting trusted public-sector AI solutions.

Earning trust: Public communication and accountability

Technology alone will not guarantee success. The legitimacy of AI deployment in public services depends on a strong foundation of public trust. Singapore’s proactive approach to workforce development already provides a powerful blueprint for demonstrating how automation can enhance jobs rather than replace them. For example, the successful retention of Singaporean crane operators to remotely supervise multiple cranes and vehicles has increased productivity while improving work quality, serving as a model for how AI can complement human skills.

Trust also hinges on responsible governance. Clear and accessible frameworks that explain how personal data is collected, used, and protected will help citizens feel secure. Regulatory sandboxes and test beds enable innovation to be trialled under controlled conditions, accelerating time-to-market for safe, ethical solutions. This pragmatic approach will guide AI oversight, balancing innovation with necessary guardrails and addressing key concerns around consumer protection and privacy. Concurrently, cybersecurity protocols must evolve continuously to keep pace with emerging threats, particularly as AI becomes increasingly intertwined with critical infrastructure and national security. Education completes the picture: both public officers and citizens must understand AI’s capabilities, limitations and safeguards. When this happens, trust increases, allowing room for bolder, more impactful deployments that deliver real public value.

The measure of leadership in the AI era

Singapore’s early AI deployments have demonstrated a national willingness to innovate and govern responsibly. But leadership will not be measured by how early programmes are launched; it will be determined by how effectively they scale into systematic impact. This means eliminating structural bottlenecks, modernizing IT infrastructure, forging deeper public–private alliances and building a culture of trust that supports adoption.

Achieving this will deliver dividends in the form of faster services, smarter use of resources, and a deeper, more transparent relationship between the government and its citizens. The pilots have shown what AI can achieve in controlled contexts. Now is the moment to prove what it can deliver at the national scale, establishing Singapore not only as a pioneer but as the benchmark for AI-enabled government worldwide.

The views and opinions expressed in this article are those of the author and do not necessarily reflect those of CDOTrends. Image credit: iStockphoto/Sashkinw