Bobby Wee (BW): <\/strong>Right now, the most consistent, provable ROI is still cost and productivity, but the mix is shifting. <\/p>\nOn the cost reduction and efficiency side, the \u201cfirst dividend\u201d is coming from automating Tier-1 support, accelerating software delivery, reducing manual compliance effort, improving incident response, and optimising supply chain and forecasting. These gains are measurable in weeks: faster cycle times, fewer escalations, lower cost to serve.<\/p>\n
The \u201csecond dividend\u201d, revenue generation, is emerging through AI-enabled product features, personalisation at scale, faster time to market, and new premium tiers such as AI copilots bundled into enterprise services. Industry-specific models that create defensible intellectual property are also beginning to appear.<\/p>\n
What has changed since last year is confidence. In 2024 to 2025, many pilots proved technical feasibility. In 2026, boards want unit economics: cost per task, cost per resolved ticket, cost per code change, conversion uplift \u2013 hard numbers.<\/p>\n
The positive macro signal is that Singapore is doubling down on capability building, with major public investment into AI research announced through 2030, including support for responsible and resource-efficient AI and talent development. That kind of long-horizon commitment tends to unlock private-sector adoption because companies know the ecosystem will be there.<\/p>\n
Thomas Laboulle (TL): <\/strong>The most tangible value in customer experience today is still efficiency, but the context has evolved. <\/p>\nLast year, many pilots focused on demonstrating what AI could do. This year, the focus has shifted to whether those capabilities can be sustained in production. Boards and executive teams are asking a more fundamental question: does this system reliably reduce operational load without introducing risk?<\/p>\n
In enterprise customer experience operations, efficiency gains remain the most immediate and measurable source of ROI. Improvements in automation rates, first-contact resolution and after-call work show up quickly in operational metrics. Importantly, these gains only matter if they persist beyond pilot environments.<\/p>\n
Revenue impact absolutely matters, and it is emerging through higher retention, faster resolution and more consistent service quality. However, those benefits tend to compound over time. Right now, enterprises are prioritising solutions that can move from experimentation to dependable, day-to-day operations.<\/p>\n
Praveen Raina (PR): <\/strong>Banking is both a business of scale and trust, and our use of AI is delivering value on both fronts. For our customers, AI is enabling a new level of hyper-personalisation. <\/p>\nWe are shifting from reactive servicing to proactive, value-adding engagement and product customisation that deepens relationships and strengthens customer trust. That increased relevance is already translating into measurable economic value.<\/p>\n
The operational returns are equally pronounced. Intelligent document processing is shortening turnaround times, while AI-enabled engineering tools have reduced coding and testing effort by 20 to 30 per cent, enabling us to deliver faster and more consistently. <\/p>\n
AI-driven anomaly detection and accelerated response times are also strengthening system resilience, ensuring our platforms remain stable and reliable.<\/p>\n
The buzzword for 2026 is \u201cAgentic AI\u201d \u2013 systems that don\u2019t just summarise text but actively execute workflows. From each of your specific vantage point, are we actually seeing this shift in deployment yet, or are enterprises still largely stuck in the \u201cchatbot\u201d phase?<\/strong><\/p>\nTL: <\/strong>We are seeing a shift, but it is more subtle than the term \u201cagentic AI\u201d suggests. In customer experience, most enterprises have moved beyond basic chatbots conceptually. The challenge has been translating that ambition into production systems that can operate reliably at scale. <\/p>\nWhat we see today is not widespread deployment of fully autonomous agents, but rather carefully scoped systems that can execute specific tasks within defined boundaries.<\/p>\n
The core issue is not intelligence, but control. Many agentic systems prioritise flexibility and autonomy, assuming that better prompting or reasoning will keep AI aligned. In enterprise customer experience, that assumption breaks down quickly. <\/p>\n
When AI is allowed to interpret processes instead of follow them, it introduces what we call \u201cprocess hallucinations\u201d: skipping mandatory steps, deviating from approved workflows or exceeding authority. This is distinct from the well-known problem of AI generating incorrect text. <\/p>\n
Process hallucination occurs when an AI agent confidently executes the wrong sequence of actions, or claims to have completed a step it never performed. In multi-step workflows, even small errors compound rapidly.<\/p>\n
In regulated environments, enterprises and government agencies cannot afford AI systems that interpret processes freely. As a result, the most successful deployments today are those where autonomy is introduced incrementally, with clear guardrails and escalation paths.<\/p>\n
So while the direction is clear, the reality is that enterprise customer experience is progressing through controlled, supervised autonomy rather than a sudden leap to fully agentic systems. <\/p>\n
There is also a significant amount of what the industry is beginning to call \u201cagent washing\u201d, where existing chatbots and robotic process automation tools are rebranded as agentic AI without any meaningful change in capability. Enterprises should look past the labels and ask a simple question: does this system follow governed processes, or does it improvise?<\/p>\n
PR: <\/strong>We are already seeing the shift. Most banks are moving well beyond the \u201cchatbot\u201d phase to piloting or deploying AI that orchestrates and executes multi-step workflows. <\/p>\nAt OCBC, we treat agentic AI not as a buzzword, but with a deliberate and disciplined approach to embedding it across our technology stack. <\/p>\n
Autonomous agents are already supporting areas such as automated Know-Your-Customer (KYC) due diligence, where the system actively assesses the legitimacy of clients\u2019 wealth and transactions, and relationship managers review and refine the final output.<\/p>\n
As we move from passive assistance to active execution, success hinges not on plug-and-play solutions but on deep integration with enterprise systems, strong governance and coherent orchestration of AI, digital capabilities and data with our people. <\/p>\n
This is how we enhance customer experience, scale customer acquisition and ensure our customers remain protected.<\/p>\n
\n
Racks Central operates a Tier 3-ready data centre purpose-built to support the mission-critical requirements of enterprise clients across the region. PHOTO: RACKS CENTRAL<\/span><\/figcaption><\/figure>\n<\/p><\/div>\nBW: <\/strong>We\u2019re absolutely seeing the shift, but it\u2019s uneven. The \u201cchatbot phase\u201d was about answering; the agentic phase is about doing \u2013 drafting a proposal, pulling data from systems, creating tickets, running checks, pushing a change end to end. <\/p>\nIn production today, the best deployments are bounded agents \u2013 <\/strong><\/em>narrow scopes, clear permissions, strong identity controls and an auditable trail. <\/p>\nEnterprises are moving fastest in rules-heavy repetitive workflows: IT operations, customer support triage, finance operations, developer workflows and network operations. The reason is simple: you can constrain the agent, measure outcomes and roll back safely.<\/p>\n
From Singapore\u2019s ecosystem standpoint, what\u2019s exciting is that governance is starting to catch up to capability. The Infocomm Media Development Authority (IMDA) recently launched a Model AI Governance Framework for Agentic AI, and that is a strong signal that we\u2019re moving from \u201cwow demos\u201d to responsible deployment at scale.<\/p>\n
As AI models become more autonomous, the \u201cblack box\u201d problem grows. How do we balance the need for advanced, autonomous AI with the strict requirement for explainability \u2013 especially in highly regulated sectors like finance and telecoms?<\/strong><\/p>\nBW: <\/strong>We shouldn\u2019t treat explainability as \u201call or nothing\u201d. The practical approach is risk-tiering. For low-risk use cases such as marketing copy drafts and internal knowledge search, allow higher autonomy with monitoring. <\/p>\nFor medium-risk applications such as service recommendations and operational decisions, policy constraints, evaluation and human review thresholds are required. <\/p>\n
For high-risk decisions involving credit, fraud, KYC or telco critical network changes, the requirement must be full traceability, documentation and human accountability, with human-in-the-loop approval at the final stage.<\/p>\n
In finance, Singapore has already put strong principles in place. The Monetary Authority of Singapore\u2019s Feat Principles are a good reference point: fairness, ethics, accountability and transparency are exactly the scaffolding you need when models get more powerful. <\/p>\n
On the enterprise side, assurance tooling is maturing too. AI Verify Foundation and IMDA\u2019s work around AI testing and assurance \u2013 including pilots aimed at codifying norms for technical testing \u2013 helps move governance from \u201cpolicy statements\u201d to repeatable engineering practice.<\/p>\n
My view is that the winners will be those who treat governance like cybersecurity: designed in, continuously tested and operationally owned.<\/p>\n
TL: <\/strong>The key is to stop treating explainability as a model problem and start treating it as a systems problem. <\/p>\nIn regulated environments, explainability does not come from exposing an AI\u2019s internal reasoning, but from ensuring that every action follows a deterministic, auditable process. From an enterprise perspective, what matters is knowing which steps were followed, which rules applied, and why an action was permitted or escalated. <\/p>\n
This is why flexibility-first AI frameworks struggle in production. When process control is handled through prompts or emergent behaviour, outcomes cannot be guaranteed or reproduced. In contrast, enterprise-grade systems must encode compliance, authority boundaries and mandatory steps directly into the architecture.<\/p>\n
In customer experience, this means separating conversational flexibility from execution logic. AI can engage naturally with customers, but actions that carry risk must be governed by explicit process controls rather than inferred behaviour. When those controls are designed into the system, explainability becomes a property of the architecture, not the model.<\/p>\n
This approach allows organisations to benefit from advanced AI capabilities while maintaining the level of accountability that regulators and customers expect.<\/p>\n
The regulatory environment is reinforcing this direction. <\/p>\n
In Singapore, MAS published its Consultation Paper on Guidelines on AI Risk Management in November 2025, setting expectations for financial institutions to ensure transparency and explainability proportionate to each system\u2019s risk and impact. IMDA launched its new Model AI Governance Framework for Agentic AI in January 2026. <\/p>\n
In Europe, full enforcement of the EU AI Act\u2019s high-risk provisions for new systems begins on Aug 2. And ISO\/IEC 42001 is established as a certifiable global standard for AI management systems. <\/p>\n
For enterprises like Toku operating across multiple jurisdictions, the question is no longer whether to build governance into AI systems, but how quickly they can do so.<\/p>\n
PR: <\/strong>In a highly regulated financial services sector, \u201ctrusted autonomy\u201d must be a prerequisite, not an afterthought. As AI agents take on longer and more complex workflows, the risk of opacity naturally increases.<\/p>\nWe address this by embedding explicit validation checkpoints with strong guardrails directly into the design of every AI workflow. Every autonomous action is underpinned by OCBC\u2019s AI governance framework, ensuring that even the most advanced models remain auditable, explainable and compliant with financial standards.<\/p>\n
\n
OCBC has set ambitious targets for employee AI augmentation by 2027. PHOTO: YEN MENG JIIN, BT<\/span><\/figcaption><\/figure>\n<\/p><\/div>\nOCBC has set ambitious targets for employee AI augmentation by 2027. As it moves from internal tools to customer-facing agentic tools, how does the bank manage the risk of an AI agent making a financial decision or recommendation without human oversight?<\/strong><\/p>\nPR: <\/strong>Our philosophy is that AI should augment human judgement, not replace accountability. Even as we scale AI across the organisation, human-in-the-loop architecture remains firmly embedded in any decision affecting customers\u2019 financial outcomes. <\/p>\nAI provides the speed, scale and intelligence to surface insights, but the final ethical and financial mandate rests with our people. This balance preserves the standards of care and trust that are fundamental to us as a bank.<\/p>\n
We see a tension between \u201cGreen AI\u201d (sustainability) and the massive compute power required for generative AI. As a bank committed to sustainability goals, how does OCBC reconcile the energy footprint of training or running these massive models with your ESG commitments?<\/strong><\/p>\nPR: <\/strong>As AI adoption accelerates, the focus must shift from scaling to deploying AI responsibly and efficiently. At OCBC, we approach this through enhancing the energy efficiency of our infrastructure and using technology itself to further optimise energy consumption.<\/p>\nOur data centres currently run on innovative cooling technologies at the server level, which drastically lower the energy footprint of our models. In parallel, we are partnering institutes of higher learning to explore how AI, machine learning and Internet of Things can dynamically monitor and optimise energy consumption.\u00a0<\/p>\n
For us, digital innovation must advance hand in hand with our sustainability targets, not come at their expense.<\/p>\n
\n
Racks\u00a0Central\u2019s construction site of a data centre at Iskandar Halal Park in Pasir Gudang, Johor. It is building three data centres on two plots of land spanning roughly the size of 10 football fields. PHOTO: HARITH MUSTAFFA<\/span><\/figcaption><\/figure>\n<\/p><\/div>\nAI racks are now pushing power densities of 50 to 100 kilowatt, far beyond traditional limits. With the Singapore-Batam digital corridor becoming critical, how is the physical infrastructure evolving to handle this heat <\/strong>\u2013 and are we moving fast enough towards liquid cooling?<\/strong><\/p>\nBW: <\/strong>High-density AI is forcing a complete redesign of the data centre as a thermal and electrical machine asset, not merely real estate. <\/p>\nAt 50 to 100 kW per rack, air cooling alone becomes increasingly inefficient or space-intensive. The industry is moving towards a spectrum: enhanced air plus containment for moderate densities, rear-door heat exchangers, direct liquid cooling for high density and immersion for specialised deployments.<\/p>\n
Are we moving fast enough? We\u2019re moving because we must. The Singapore-Johor-Batam digital corridor, supported by emerging Special Economic Zone frameworks, is becoming strategically critical for the region. <\/p>\n
It allows operators to combine Singapore\u2019s network density, enterprise demand and regulatory maturity with Johor and Batam\u2019s expansion runway \u2013 land, power, water availability and scale \u2013 while remaining tightly interconnected from a latency, operations and governance perspective. <\/p>\n
On the Singapore side, policy is also pushing efficiency harder. IMDA\u2019s Green Data Centre Roadmap and evolving requirements around new capacity are raising the bar on power usage effectiveness and sustainability outcomes. That is exactly what you want when AI demand is growing. <\/p>\n
The bottom line is liquid cooling is no longer a nice-to-have. It is a core competency, and the corridor strategy is how we scale responsibly.<\/p>\n
AI chips become obsolete much faster than standard servers. Does this create a new challenge for data centres in terms of frequent retrofitting or e-waste management?<\/strong><\/p>\nBW: <\/strong>Yes, graphics processing unit (GPU) obsolescence creates a challenge, but also an opportunity to modernise how data centres think about lifecycle.<\/p>\nAI clusters have shorter refresh cycles than traditional servers, and the answer is designing for modularity and circularity. That means modular power and cooling infrastructure so upgrades do not require ripping out entire rows, quick-connect liquid loops that make hardware swaps safer and faster, standardised mechanical and electrical \u201crails\u201d. <\/p>\n
New GPU generations can drop in with minimal downtime, and certified reuse and resale pathways (accompany) secure sanitisation so hardware earns a second life rather than becoming waste. The market will increasingly reward operators who can offer compute refresh without disruption and who can demonstrate a credible e-waste and carbon accounting story. <\/p>\n
The data centre of the AI era is not just built for uptime; it is built for upgrade velocity and responsible decommissioning.<\/p>\n
Toku has long emphasised that Western AI models often struggle with the linguistic fragmentation of our region. As you roll out voice AI agents, how are you solving the \u201caccent gap\u201d and hallucination risks when dealing with diverse Singlish, Manglish or Bahasa nuances in real time?<\/strong><\/p>\nTL: <\/strong>In practice, transcription accuracy is the foundation of everything that follows. In linguistically diverse regions, small errors at the speech recognition level can quickly cascade into incorrect intent detection, inappropriate responses or broken workflows. This is particularly true in real-time voice interactions, where there is little opportunity for correction. <\/p>\nReal-time voice AI in APAC must handle code-switching, local speech patterns, and telephony-grade audio without degradation. If the system mishears, no amount of downstream intelligence can recover safely. The reality is straightforward: if you are not testing for code-switching, your multilingual voice agent is not production-ready.<\/p>\n
This is precisely why we invest in proprietary transcription technology trained specifically for the linguistic complexity and audio quality conditions typical of our target markets, rather than relying on generic global models. <\/p>\n
Hallucination risk is managed in a similar way. Rather than relying on the model to infer responsibly, we constrain responses and actions through approved knowledge, defined workflows and automatic escalation when ambiguity arises. The goal is a correct, compliant customer outcome every time, even in messy, real-world conversations.<\/p>\n
In the context of customer experience, where does Toku draw the line? When must a human agent intervene, and is that handoff becoming smoother or more complex as AI becomes more confident?<\/strong><\/p>\nTL: <\/strong>Human intervention is essential whenever judgement, risk or emotional nuance exceeds what an automated system can safely handle. This includes scenarios involving financial decisions, identity verification, regulatory exceptions or heightened customer distress. Critically, these boundaries should be designed upfront and not discovered after deployment. <\/p>\nIndustry best practice now centres on five escalation triggers: emotional cues detected through sentiment analysis; explicit customer requests; AI confidence falling below defined thresholds; business rules for high-value or sensitive interactions; and conversation loop detection after repeated failed attempts.<\/p>\n
When done well, AI makes the handoff to human agents smoother by preserving context, summarising what has occurred, and routing the interaction appropriately. This allows human agents to focus on resolution rather than reconstruction. <\/p>\n
Yet seamless transitions between channels remain the exception rather than the norm across most enterprise customer experience environments, which indicates both the difficulty and the scale of the opportunity ahead. <\/p>\n
In mature customer experience deployments, AI does not replace human judgement but protects it by ensuring humans intervene exactly where they add the most value.<\/p>\n
There is growing demand for AI-driven sentiment analysis to detect customer frustration. Is there a privacy line here? How does Toku balance the benefit of empathy-driven AI with the \u201ccreepy\u201d factor of a machine analysing a customer\u2019s emotional state?<\/strong><\/p>\nTL: <\/strong>There is a clear line between using sentiment to improve service and using it in ways that undermine trust. In customer experience, sentiment analysis should function as a real-time indicator that an interaction may require additional support or escalation. It should not be used to categorise or profile customers beyond what is necessary to resolve the issue at hand. <\/p>\nResponsible use means minimising data, being transparent about analysis and embedding sentiment strictly within operational workflows.<\/p>\n
The regulatory picture is becoming clearer, which is helpful for enterprises navigating this space. The EU AI Act prohibits the use of emotion recognition systems in workplaces and educational settings, with those prohibitions applying from February 2025 and potential penalties of up to 35 million euros (S$51.8 million) or 7 per cent of global annual turnover for breaches. <\/p>\n
In Asia-Pacific, no jurisdiction has banned sentiment analysis. Singapore\u2019s approach remains voluntary and principles-based, and IMDA is backing efforts like the MERaLiON Consortium to advance research in multilingual and empathetic AI, including emotion recognition across South-east Asian languages. <\/p>\n
The key distinction for enterprises is purpose: sentiment used to improve service quality and trigger appropriate escalation is a very different proposition from sentiment used to profile or manipulate customers. <\/p>\n
Trust is central to customer experience, and any use of AI must reinforce \u2013 not weaken \u2013 that trust.<\/p>\n
Last but not least, if we convene this roundtable again in 18 months, what is the one AI topic we are currently obsessed with that will be obsolete?<\/strong><\/p>\nPR: <\/strong>The conversation will increasingly shift from whether AI systems can autonomously complete tasks to how organisations orchestrate workflow and workforce around them. <\/p>\nThis next phase is about redesigning the very fabric of the bank so that humans and autonomous agents operate in a secure, seamless tandem. At the same time, the pace of AI evolution is outstripping traditional governance. <\/p>\n
The challenge will be to build agile governance and operating structures capable of keeping pace. Organisations will need to manage the ever-evolving autonomous systems in real time, capturing new value while maintaining a strong and consistent risk posture.<\/p>\n
TL:<\/strong> I expect the debate around \u201cchatbots versus agents\u201d will feel dated, and agentic capability will increasingly be assumed. But I would go further: the notion that AI can wholesale replace human agents in customer service will also look dated. <\/p>\nVery few enterprises have actually reduced service headcount because of AI, and some that moved too aggressively are already reversing course. The vision of fully agentless customer service will be quietly retired.<\/p>\n
What will replace it is a more grounded conversation about production readiness. The real work involves mapping processes, establishing governance and building operational foundations that rarely make headlines. That is not glamorous, but it is where lasting value is created.<\/p>\n
In customer experience, the conversation will shift away from autonomy as a headline feature and towards production readiness: accuracy, control, compliance, and operational resilience. Those are the factors that ultimately determine whether AI delivers lasting value. <\/p>\n
The companies that win will not be those with the most impressive demos, but those that can deploy reliably at scale, in production, in regulated environments, across multiple languages and explain to auditors exactly how their AI makes decisions.<\/p>\n
BW: <\/strong>The topic that will feel obsolete is the obsession over \u201cwhich model is best\u201d as a standalone debate \u2013 model A versus model B, parameter counts, leaderboard chasing. <\/p>\nIn 18 months, the real competitive edge will be orchestration and assurance: how enterprises manage fleets of models and agents; how they validate outputs continuously; how they prevent agent-driven security incidents; and how they govern autonomy across vendors, jurisdictions and critical systems. <\/p>\n
The new challenge that replaces it is the reliability and security of autonomous workflows at scale (agentic cyber risk), permissioning, auditability and resilience, alongside the physical reality of power, cooling and grid integration to sustain AI growth. <\/p>\n
If we get that right, Singapore can be more than a user of AI \u2013 we can be the region\u2019s reference point for trusted AI infrastructure, supported by serious national investment and governance leadership.<\/p>\n
\n
\n
Decoding Asia newsletter: your guide to navigating Asia in a new global order. Sign up here to get Decoding Asia newsletter.<\/span> Delivered to your inbox. Free.<\/p>\n<\/div>\n<\/div>\n<\/div>\nRead Full Article At Source <\/a><\/p>\n","protected":false},"excerpt":{"rendered":"Three business leaders from banking, customer experience and data centre infrastructure share what it actually takes to make AI work reliably, responsibly and at scale…<\/p>\n","protected":false},"author":1,"featured_media":1864,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"fifu_image_url":"","fifu_image_alt":"","footnotes":""},"categories":[2611],"tags":[],"class_list":["post-40306","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-buzz-headlines","wpcat-2611-id"],"_links":{"self":[{"href":"https:\/\/sgbuzz.com\/index.php?rest_route=\/wp\/v2\/posts\/40306","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/sgbuzz.com\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/sgbuzz.com\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/sgbuzz.com\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/sgbuzz.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=40306"}],"version-history":[{"count":0,"href":"https:\/\/sgbuzz.com\/index.php?rest_route=\/wp\/v2\/posts\/40306\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/sgbuzz.com\/index.php?rest_route=\/"}],"wp:attachment":[{"href":"https:\/\/sgbuzz.com\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=40306"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/sgbuzz.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=40306"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/sgbuzz.com\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=40306"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}
![\"\"](\"https:\/\/cleaver.cue.sph.com.sg\/public\/businesstimes\/incoming\/rp8qzt-dtroundtable30jpg\/alternates\/BASE_PORTRAIT\/dtroundtable30.jpg\")