Holistic AI agent visibility will become a business imperative
Over the next year, organisations will realise they need visibility into the agents running across their entire network, as teams spin up AI systems from development tools, cloud platforms, and countless other sources, often without centralised oversight. Agent platforms that can discover and catalog these distributed AI systems will emerge as the clear winners in the enterprise market.
As agents increase system usage and computing costs, organisations will demand precise ROI tracking and qualification for their AI investments. Companies will stop treating agent deployments as untracked experiments and will start requiring the same financial accountability they expect from any other enterprise technology. The most successful organisations will implement agent discovery platforms that provide visibility into what agents are running, the resources they’re consuming, and whether they’re delivering measurable business value.
Human-centric only identity systems will fail in an agent-to-agent world
Over the next year, organisations will confront an access and permissions crisis as agent-to-agent interactions expose the limitations of traditional access control systems. Unlike human users or simple automations, agentic AI systems communicate with each other, delegate tasks, and make decisions that cascade across multiple systems, exposing gaps in traditional composite identity solutions.
When one agent instructs another, existing permission frameworks break down because they were designed for individual human actors, not autonomous systems. Organisations that continue using human-centric permission models will find themselves unable to trace decision-making chains, audit agent actions, or maintain security as their AI systems become increasingly interconnected and autonomous.
Security teams will prioritise clearly defined, high-impact AI use cases
AI has already proven to reduce the rate of false positives and help streamline security operations. Successful implementations start with clearly defined, high-impact use cases, such as log analysis that would overwhelm human analysts, network pattern recognition for novel threats, vulnerability prioritisation, and automated incident triage that reduces alert fatigue.
Security’s first priority should be to document institutional knowledge across their department. AI agents need clear direction. Without company-specific context, they will only deliver technical debt. This documentation will also help to strengthen and standardise internal security processes.
Strategic AI-human collaboration will define competitive advantage in 2026
The winners won’t be the companies that adopt AI fastest. They’ll be the ones who are most intentional about what they assign to AI versus humans. With 90 % of executives expecting agentic AI to become standard within three years (according to GitLab’s The Economics of Software Innovation report), the real differentiator will be knowing exactly which tasks benefit from human creativity and judgment versus which should be automated.
Organisations that are thoughtful about this calibration will create compounding advantages, freeing developers to focus on high-value architectural decisions and strategic thinking, while AI handles code generation and routine maintenance. Companies that get the balance wrong face a double penalty: human talent wasted on automatable work, and AI making decisions requiring nuanced judgment.
– Craig Nielsen, Vice President, Asia-Pacific & Japan, GitLab
————————————————————————————————————
2026 will be the year AI becomes a discipline, not an experiment
2025 was the year artificial intelligence became tangible. Data centres expanded across Southeast Asia, GPU procurement accelerated and AI moved from concept to operational deployment. But as adoption scaled, the constraints also became visible: power availability, operational cost, data management maturity and regulatory alignment. These realities are now shaping enterprise investment and national digital infrastructure policy.
For the past two years, many organisations approached AI as primarily a compute challenge. The assumption was that more accelerators would drive more capability. The surprise is that the limiting factor is no longer the model or the hardware. It is the ability to move, govern and secure data at scale. The true constraint is the data pipeline, the storage and network architecture that support it, and the operational discipline required to run these systems reliably.
The next phase of AI will be agentic, with systems making decisions and initiating actions in financial services, healthcare, logistics and manufacturing. For governments and regulated industries, this raises new questions of accountability, transparency and security. These systems require data that is accurate, governed and available in the right context, as well as infrastructure designed for high-throughput training, low-latency inference and continuous resilience.
This is where the conflict is emerging: regional demand for AI capability is rising, but the physical capacity to support it is finite. Energy efficiency is becoming a strategic concern. Data centre expansion in ASEAN — particularly in Singapore, Malaysia, Indonesia, Thailand, and the Philippines — is now constrained by power and land availability. The region cannot simply scale capacity endlessly. Modernising storage and compute platforms to dense, efficient architectures is now essential to ensuring that AI is economically and environmentally sustainable.
Data sovereignty is accelerating for similar reasons. Nations are strengthening expectations around residency, operational control and cybersecurity assurance. The future will not be a choice between public cloud and on-premises infrastructure, but hybrid models that combine local control with secure data mobility. What happens in Southeast Asia will matter globally, because the region is becoming a test case for how economies balance AI growth, sustainability and sovereignty.
The organisations and nations that lead in 2026 will treat data as a strategic asset. They will invest in energy-aware architecture, resilient data platforms and continuous talent development. AI maturity will not be defined by who has the most compute, but by who can operate the most disciplined and efficient data ecosystem.
– By Lawrence Yeo, ASEAN Solutions Director, Hitachi Vantara
————————————————————————————————————
AI empowerment wins over AI replacement.
Do you want to play to win, or not to lose? In 2026, leaders will face a spectrum between two options: either use AI to eliminate jobs or use AI to empower people to create a competitive advantage. It is becoming increasingly clear that AI should empower people – not replace them – and companies will need bold and inspirational leaders to invest in their workforce through continued change.
– Bryan Harris, Chief Technology Officer, SAS
CIO? That’s Chief Integration Officer to you.
In 2026, CIOs will answer the call to orchestrate the agentic AI future. As AI agents proliferate, the CIO’s role will decisively shift from tech enabler to ecosystem integrator: the Chief Integration Officer. AI governance, integration and cross-functional leadership will shape the day of every CIO as we determine the future of IT architecture in an agent-led world.
– Jay Upchurch, Chief Information Officer, SAS
Agentic AI matures.
By 2026, agentic AI – systems that act, decide and adapt autonomously – will move from pilots to the core of how organisations operate and serve customers. Those investing in the right infrastructure, governance and skills will unlock smarter decisions and seamless experiences. Those who don’t will fall behind both in performance and in meeting customer expectations.
– Jennifer Chase, Chief Marketing Officer, SAS
HR’s new workforce: humans + agents.
By 2026, HR leaders will manage more than people, they’ll manage AI agents too. As agentic AI becomes part of daily workflows, HR will define new policies for onboarding, performance and collaboration between humans and digital coworkers. The future of workforce management will be hybrid-human and machine.
– Jenn Mann, Chief Human Resources Officer, SAS
Data centre downfall.
Major investments in data centre buildouts will prove impractical as costs come home to roost; expectations were high, but resulting revenue wasn’t enough to cover the expense. Tech companies angle for alternatives. Economics experts crow told-you-so.
– Jared Peterson, Senior Vice President, Platform Engineering, SAS
Meet your new coworkers: Agentic AI.
As we mark half a century of innovation, resilience and growth, 2026 ushers in a new era of enterprises evolving into ecosystems where AI agents are no longer tools; they are teammates. Enterprises will be expected to operate with mixed human-AI teams, where agents act as trusted collaborators, executing tasks, sharing context and learning continuously alongside people.
– Udo Sglavo, Vice President, Applied AI and Modelling Research & Development, SAS
Trust and innovation couple up.
In 2026, the AI debate will no longer be one of innovation versus trust. As government regulation of AI remains inconsistent, corporate self-governance will extend to include the necessary guardrails to enable AI in the enterprise responsibly. The organisations that thrive won’t simply be those that deploy AI first; it will be those that recognise the strategic reality that governance isn’t a restraint on innovation, it’s a necessary companion.
– Reggie Townsend, Vice President, Data Ethics Practice, SAS
————————————————————————————————————
Prediction 1: CFOs will evaluate AI more strategically.
In 2026, CFOs will increasingly favor reliable, rules-based automation over AI-everywhere approaches. The real shift is toward tools that deliver clarity and accuracy without unnecessary complexity. The trend to watch: solutions that blend automation with selective, purposeful AI, proving that you don’t need AI everywhere to move faster and get revenue right.
Jagan Reddy, founder & CEO, RightRev
————————————————————————————————————
Building the Intelligent, Connected Enterprise
AI adoption has surged globally, with trillions invested in advanced tools. Yet many organisations still struggle to see measurable gains. Genuine productivity in an AI-driven workplace equals outcomes. But how do we distinguish between efficiency and illusion?
Fewer than 3 in 10 CEOs were satisfied with their AI results, according to Gartner. Many companies have rushed to deploy tools, achieving small process improvements but few strategic returns. This happens when adoption is fragmented from core business objectives. Productivity becomes real only when AI strengthens existing workflows and supports human decision-making.
A recent internal initiative offered one illustration. The organisation began the year by training every employee on AI, and by May, essentially all teams were using it. The real progress came not from technology alone but from culture. People were encouraged to experiment, share ideas, and apply AI to practical challenges. This collective curiosity created measurable improvements in efficiency and innovation across teams.
The fallacy of efficiency appears when AI activity increases, but outcomes do not. Real efficiency shortens decision cycles, improves accuracy, and creates new value. Strong data foundations and an integrated operating model allow AI to amplify work instead of adding noise. The goal for 2026 is to embed intelligence into business systems so that it drives measurable, sustainable growth.
Trust will also define competitiveness. As data moves freely across organisations and borders, companies must maintain confidence in how it is managed and protected. Compliance, security, and governance are the foundation of digital credibility. Businesses that uphold transparency while innovating responsibly will secure long-term advantage.
Southeast Asia illustrates both the potential and responsibility of this new phase. The region’s young, tech-literate population and fewer legacy systems make it ideal for rapid AI experimentation. But progress must be grounded in structured data, clear governance, and strong ethics. Many local businesses still rely on manual methods, so digital readiness must be built. Shared standards, responsible AI frameworks, and transparent data practices will ensure that innovation produces real value.
Talent development will be another priority. Human creativity and judgment will become central to value creation. Companies must invest in data literacy, systems thinking, and cross-functional problem-solving. Training should be continuous and tied directly to how people work with AI.
The workplace of 2026 will belong to organisations that align human potential with intelligent technology. The most successful will treat AI as a collaborator that amplifies creativity and sustainable progress.
The broader mission for organisations in this space is to transform encounters, data, and ideas into lasting innovation, and to help shape the intelligent, connected enterprises that will define the future of work.
– Kazunori Fukuda, Managing Director, Sansan Global (Thailand)
————————————————————————————————————
Cross-Market Forces Shaping APJ’s Emergence in the Global AI Landscape
APJ Takes Off as the AI Launchpad of the World
In 2026, APJ is poised to export groundbreaking AI innovation worldwide, marking a definitive maturation of its role in the global technology landscape. This signals a shift in the region from being primarily an adopter of global solutions to becoming a developer and exporter of technology.
APJ’s rise in the global AI landscape can be attributed to the collective strengths of its diverse markets. India is establishing itself as the world’s R&D engine, powered by a vast developer base and a growing network of over 1,950 global capability centres (GCCs). Previously functioning as support centres for back-office processes, GCCs are transforming into strategic hubs that drive innovation, research, and digital transformation. Many are now responsible for building and exporting global AI solutions, solidifying India’s pivotal role in the global tech landscape.
In Southeast Asia, agile start-up ecosystems in places like Singapore, Vietnam, and Indonesia are serving as real-world sandboxes for AI-first applications. This fast-paced, dynamic environment is fueling rapid innovation, reshaping diverse industries with AI-driven solutions. Meanwhile, Australia is establishing itself as a valuable testbed for developing and scaling AI solutions across sectors, while China’s leadership in AI patents, Taiwan’s semiconductor chips dominance, and Japan’s advanced R&D infrastructure are anchoring innovation at a global scale. These established powerhouses provide the deep expertise and resources needed to push the boundaries of AI technology.
From Hype to Value: AI Investments Demand Measurable Returns
APJ organisations are intensifying AI investments, with over 50% reallocating funds from other areas to double down on AI. For those planning to invest in AI agents in the next two years, close to a third (29%) already have an established investment plan. This is a deliberate, strategic decision to fund future growth by shifting away from traditional investments.
However, organisations now expect AI to deliver concrete, measurable results that will secure their position in a competitive future. Across APJ, C-suites are demanding a 2-4x return on investment within 12 to 18 months of deployment. This is driving a new level of rigor and accountability in every step of the AI project lifecycle.
– UiPath
————————————————————————————————————
AI plumbing. As the models, frameworks, and standards change, many solutions that have already been built will quickly become outdated, unsupported, or insecure. There will be a significant effort to refactor these pilots which will impede new AI development. Organisations will need to pause to work on the plumbing to define and build the new frameworks and standards needed for their organisations that are scalable, flexible, and secure.
Employees will suffer from change management fatigue. As employees adapt to a new way of working and potentially new roles, some will suffer from Germane Cognitive Load exhaustion (the mental effort required to process new information by integrating it with existing knowledge which promotes deep learning and understanding).
The compounding curve. The gap in productivity between the innovators and early adopters and the majority and laggards will quickly expand. The innovators and early adopters will become 10x’ers and organisations will need to come up with innovative ways to foster growth and learning to bring the rest of the organisation along.
Smaller is better. Models will continue to get larger and more capable, but organisations will find the best value with smaller, tailored, more cost-effective models that they can deploy internally and at the edge.
The Hype is real and there is reality in the Hype. Generative AI disillusionment accelerates as there is more awareness of AI’s influence on the news cycle through LLM generated audio and video content. The use of AI companions, the saturation of AI content in social media, and adult use of LLM’s all result in more calls for regulation. Societal blowback of AI increases as the mid-term elections approach in the US.
Integrate or get left behind. Market disruption will begin with companies that offer non-integrated products. Offerings that lack integration and data/semantic “secret sauce” will be most at risk. Integrated platforms and solutions will gain share by unlocking the value of existing customer data across products.
AI hype fades, trust endures. There will be numerous “News at 11” moments with AI as companies rush to release AI capabilities within their organisations. This will lead to an enhanced focus on companies prioritising security, fairness, accountability, and transparency. Trusted/Responsible AI becomes a competitive advantage.
– Ed Keisling, Chief AI Officer
What’s next in AI in cybersecurity
Features fueled by large language models will become table stakes AI-powered search, built-in chatbots or assistants will become standard features that users expect rather than a competitive differentiation or advantage. Products and vendors that cannot cope with this expectation will naturally lose their market position.
“Fear of missing out” will fade away
Companies will return to traditional use-case validation and rigorous return on investment metrics when investing in AI-powered solutions. This will lead to differentiation between solid AI-powered tools able to solve real use cases from plain chat wrappers.
– Pavel Minarik, VP, Product Security
————————————————————————————————————
As we enter 2026, the financial services industry stands at an inflection point. The conversation has moved beyond digital transformation to cognitive banking. Institutions are now focused on applying AI and automation to sharpen decision-making and deliver more meaningful value to customers.
Here are three trends that will shape banking and financial services in the year ahead.
- AI becomes a business partner, not just a tool
AI should be applied from the core of financial institutions, supporting decision-making across credit assessment, fraud detection, and operational efficiency from the core since that’s where the centre of deep banking knowledge resides. Rather than isolated use cases, we will see AI embedded in workflows, acting as a trusted co-pilot for relationship managers, risk teams, and customer service operations. Success will depend on having an intelligent platform to work with data foundations to ensure seamless workflows, processes, transparency and accountability with human oversight. - The shift to intelligent banking will accelerate
Banks are moving beyond basic digitisation toward AI-enabled, real-time decisioning across credit, fraud, risk, and customer engagement. What used to take days will happen in seconds. This will redefine operational efficiency and customer expectations. Core platforms will need to be ready to support streaming data, embedded AI models, and instant processing which is why our latest hybrid and cloud-native architectures will become non-negotiable. - Smarter automation in the back office
Automation will evolve from task-level optimisation to intelligent orchestration. Financial institutions will use AI-driven workflows to reduce reconciliation time, simplify reporting, and strengthen compliance oversight. The focus will shift from cost reduction to accuracy, scalability, and freeing human talent for higher-value work.
In the year ahead, technology will not just support financial services but define its competitive edge. The banks that invest strategically in intelligence, trust, and resilience today will set the benchmark for the industry’s next phase of growth.
– Cassandra Goh, Group CEO, Silverlake
————————————————————————————————————
By 2026, we’ll see a clear divide, not between companies that use AI and those that don’t, but between those treating AI as a cost-cutting exercise and those strategically redesigning their customer experience around AI. The best organisations will not use AI to cut costs – they’ll use it to deliver greater value to customers by better understanding their needs at every single moment and touchpoint.
Our recent 2026 Consumer Experience Trends report found that over half (55%) of consumers in Singapore had concerns about the lack of human connection with automated interactions. This clearly proves there’s very much still a role for human led interaction, even in an AI savvy nation like Singapore.
Singapore is one of the world’s most AI-forward markets, yet even in this advanced society, only 40% of people trust organisations to use AI responsibly. This low trust level presents a real challenge for businesses relying on AI to transform consumer experience. So, as we head into 2026, we need to focus on using AI to equip humans with better insights and faster solutions, rather than replacing them entirely.
By late 2026, I hope we’ll see the emergence of a new customer experience operating model, one where AI will allow organisations to better serve customers by handling transactional tasks so that teams can focus on more complex interactions. These organisations will be able to scale premium service experiences, without scaling costs. But this won’t happen overnight, and it won’t happen without trust. Organisations that empower and trust their employees to use AI ethically will get the best results. They will have a workforce that is confident and ready to tackle high-stake interactions.
Beyond AI, leading organisations will focus on hyper-personalisation in real time. They’ll achieve this by understanding what customers are doing and where they are (behavioural and contextual signals), and connecting all the scattered pieces of information from every touchpoint, whether it’s calls, chats, reviews, social media or transactions.
– Irene Ng, Head of CX Solution Strategy, SEA at Qualtrics
————————————————————————————————————
Before the AI bubble pops, you can be ready ahead of the game.
Whether the AI bubble bursts in 2026 or 2027, forward-looking teams can prepare now for the opportunities that follow. As technologists, we’ve seen this pattern before: Once the hype cools, infrastructure becomes dramatically more accessible. That means more GPUs, more distributed data centres, more storage, more electricity, more edge processing and more capacity overall, all at price points that open new doors for innovation. Smart organisations will treat this as a moment to plan, so they’re ready to capitalise on the windfall on the other side of the bubble.
We expect this GPU surplus to accelerate what our database can deliver. We see a future of incredibly fast, massively parallel processing for operational workloads, differentiated data analysis capabilities, highly scalable vector indexing and search, and new ways to serve AI-powered applications at global scale. With GPUs, memory, storage and networking becoming cheaper and more widespread, the price-performance curve shifts in favor of builders. And we couldn’t be more excited about what that unlocks.
The next-gen of developers will act more like conductors guiding fast-moving teams.
The day-to-day work of a developer is shifting. Developers who want to stay ahead will use AI the way a head chef runs a busy kitchen, directing parallel tasks, comparing multiple options, deciding what’s worth keeping and pushing work forward quickly. The real skill is orchestration, not trying to personally hand-craft every line of code. That shift will help teams ship faster and stay relevant.
The biggest advantage will come from understanding the higher levels of the system: how data flows, how subsystems behave under load and how to keep the bigger picture in focus across an increasingly distributed world that spans edge and cloud. Developer data platforms that support quick iteration, flexible data models and reliable edge-to-cloud performance will give teams what they need to supervise and collaborate with AI so they can move faster than the competition.
– Steve Yen, Co-Founder, Couchbase
————————————————————————————————————
How AI-led thinking, personalised learning, and new assessment models will redefine schooling in 2026
Across schools in Singapore and beyond, AI is reshaping how young people learn, question, and make sense of the world. But the real shift coming in 2026 is not about adopting new tools but about rethinking the skills, habits, and dispositions students need to engage with AI safely, ethically, and with confidence. As educators, our responsibility is not to prepare students for technology, but for a future where critical thinking, adaptability, and sound judgement matter more than ever.
The first major change we will see is a move from teaching students how to use AI tools to teaching them how to think with AI, as it continues to evolve. What does not change is the ability to ask good questions, evaluate sources, identify bias, and recognise when an AI-generated answer may not be accurate. These are analytical skills that take time to build, and will become central to curriculum design in the years ahead. When students understand how AI works and its limitations, they become better problem-solvers, not passive consumers of machine-generated output.
The second shift will involve personalisation becoming a normal part of classroom practice. AI already gives teachers new ways to adjust reading levels, scaffold writing tasks, and support students who are learning in a second language. In 2026, the more meaningful impact will come from how learners themselves use AI to stretch their understanding. When students can receive instant feedback, test ideas safely, or iterate on their work at their own pace, they take greater ownership of their learning. This does not replace teachers but instead gives them more time to focus on what matters – guiding students to think deeply, ask better questions, and reflect on their progress.
The third transformation will appear in assessment. Much of the public conversation on AI in schools still revolves around cheating, but that misses the point. The answer is not to ban AI or to fear its use. It is to design assessments that require students to demonstrate understanding in ways AI cannot replicate. Interviews, presentations, and collaborative projects will become more common, because they reveal how a student reasons, not just the answer they produce. When assessments capture thinking rather than recall, AI becomes a tool that supports learning rather than short-circuits it.
These changes reflect a broader truth: AI will sit alongside our students for the rest of their lives. Schools must give them the mindset and maturity to use it well. That means grounding AI education in ethics, human judgement, and leadership—skills that will matter long after today’s tools are outdated. If we can help students approach AI with curiosity, responsibility, and confidence, we will prepare them not only for the technologies of 2026, but for the kind of people they will become long after they leave our classrooms.
– Justin Kirby, Senior Head of Academic Pathways & Student Achievement at XCL World Academy
————————————————————————————————————
From catching bad actors to understanding good behaviour
The implications of agentic and adversarial AI are significant. Traditional fraud prevention has been built around detection, which means spotting anomalies, scoring risk and identifying signals that don’t fit the pattern. But as agentic AI reshapes how those patterns are forged, the industry’s focus is starting to flip. 2026 won’t be about finding the bad actors, it will be about understanding what good, genuine behaviour really looks like.
The boundary between genuine and synthetic activity is blurring. Generative AI can now simulate human interaction with high accuracy, including realistic typing rhythms, believable navigation flows, and deepfake biometrics that replicate natural variance. The traditional approach of searching for the red flags no longer works when those flags can be easily fabricated.
The next evolution in fraud detection will come from baselining legitimate human behaviour. By modelling how real users act over time and looking at their rhythms, routines and inconsistencies, we can identify the subtle deviations that synthetic agents struggle to mimic. It’s the behavioural equivalent of knowing a familiar face in a crowd. Trust comes from recognition, not reaction.
– George Pace, Sr. Manager Product Marketing, SEON
————————————————————————————————————
Fragmentation and the need for unified platforms
A major challenge in the year ahead will be the rise of AI silos. As generative and agentic AI gain momentum, different departments are adopting tools independently, running proofs of concept in isolation and deploying solutions without alignment. This pattern mirrors the fragmented early days of Business Intelligence (BI), where innovation was slowed by inconsistency and lack of governance. Without a unified approach, these silos make it difficult for enterprises to maintain control, enforce standards, and scale capabilities across the organisation.
Real-world use cases for AI agents will also expand rapidly. After a year of experiments, 2026 will be the point where AI agents begin driving tangible business outcomes. Financial services firms, for instance, are applying AI agents to functions such as Source of Wealth assessments, fraud prevention, and other operational tasks. According to a recent global report from Finextra Research, 97% of financial institutions operate at least one AI or machine learning use case in production. Yet nearly half of organisations remain in the middle stage of maturity where scaling, governance, and cost control are significant challenges. The next step involves operationalising AI agents at scale by connecting them to real-time, governed data and embedding them across business workflows to unlock automation that is context-aware, traceable, and secure.
– Remus Lim, Senior Vice President, Asia-Pacific & Japan, Cloudera
————————————————————————————————————
By 2026, AI will shift from assisting to genuinely agentic. The defining change will be autonomy systems that can interpret goals, take action across complex workflows, and continuously improve based on outcomes. This rapid evolution of generative AI is transforming user interactions from simple two-way exchanges into dynamic interactions that also execute tasks, enabling agents to understand intent, pull context from across channels, and complete actions on the customer’s behalf. Organisations that benefit will be those that design operations around this agentic model, enabling experiences where conversations don’t just inform decisions, they trigger them.
This shift becomes most visible in customer experience, the space where intent, conversation, and action naturally converge. In 2026, customers will engage with AI agents that can resolve issues end-to-end: reading context, personalising in real-time, and integrating directly with back-end systems. By removing the friction that makes current experiences slow or inconsistent, these systems will move confidently from understanding to action, serving as the clearest proof point for agentic AI.
As agentic systems take over repetitive, high-volume tasks, human teams will shift toward judgement, escalation, and designing the prompts, workflows, and guardrails that shape autonomous behaviour. Organisations will need new capabilities focused not only on model performance but on maintaining the logic, policies, and governance that guide how agents operate in the real world.
A major shift will be the move beyond single-agent tools. Instead of relying on one model to handle everything, organisations will adopt specialised agents that collaborate across data validation, reasoning, execution, and compliance. This will push companies to think in terms of orchestration, designing how agents communicate, escalate, correct each other, and when they should involve a human. This coordinated multi-agent approach will become the backbone of enterprise-grade autonomy, ensuring reliability, safety, and measurable impact at scale.
– Ervin Jagatić, Business Unit Director, Infobip.
————————————————————————————————————
As large language model (LLM) usage scales in organisations, a painful reality is emerging: The cost isn’t in training models, it’s in running them. Every query, every interaction, every API call adds up. And suddenly, the AI transformation that looked promising in pilots becomes unsustainable in production.
Most organisations think they have “done AI” because they subscribed to Copilot or ChatGPT. That’s necessary but insufficient. MIT research shows 95% of generative AI pilots fail from a lack of clear business justification, but there’s a deeper issue most aren’t discussing: inferencing economics.
The battlefront has shifted
As LLMs scale, the real cost driver is inferencing – running models thousands of times daily. A complex query on a large model can cost 10-100 times more than a smaller, optimised one. Your AI budget evaporates on compute, not innovation.
This is why I have come to see successful AI as solving what I call the AI triangle simultaneously:
- Speed: Deploy before your window closes. No luxury of multi-year timelines.
- Sovereignty: Protect data that gives you competitive advantage and control your infrastructure costs.
- Transformative: Move needles that justify the spend i.e. real behaviour change, not adoption metrics.
Treat these as trade-offs and you fail. Treat them as prerequisites and you transform.
What this means for architecture
Organisations must stop defaulting to the largest model. Most use cases don’t need frontier models. Match model size to problem complexity.
Optimise for efficiency, not just capability. A smaller model running 50 times faster at one-tenth the cost often beats a larger model that is marginally more accurate.
Build inference-aware systems. RAG, caching, model routing – these are not nice-to-haves, they are economic necessities.
The 2026 reality
Winners would not need to have the biggest models – they’ll have the most cost-efficient inference architecture. They’ll deploy faster through smart partnerships. And they’ll have workforces using AI daily, not just in pilots.
The real test: can you move from broad adoption to meaningful ROI? That means going beyond generic AI assistants to models pre-trained with domain knowledge, fine-tuned for your context, and optimised for sustainable inferencing costs.
Organisations that master inferencing economics will have sustainable AI transformation. Those that don’t will scale themselves into budget crises.
The question isn’t whether to adopt AI. It’s whether you can afford to run it at scale.
– Sutowo Wong, Managing Director, AI & Data, Temus
————————————————————————————————————
AI goes truly global by becoming more local
In 2026, we will see more localisation and relevance of AI innovation to markets in Southeast Asia, with more options available to cater to different purposes. AI innovation has primarily been in English at launch, with a focus on getting the technology working well. As agentic AI becomes more mature, we anticipate that the coming year will see more investments in localising AI to reflect the unique linguistic tapestry that is ASEAN. AI models fine-tuned with regional linguistic and cultural nuances will deliver more accurate and contextually relevant responses, improving customer experience and making a true business impact in the region. We will also see a variety of LLM options emerge for local businesses, as global LLMs localise and more regional, local, and even industry-specific small language models emerge to cater to varying needs.
The availability of local and region-specific AI unlocks a huge opportunity in ASEAN, empowering businesses to solve uniquely local problems. For example, more AI platforms are beginning to support Southeast Asian languages such as Tagalog, Thai, Vietnamese, Bahasa Melayu, and Bahasa Indonesia, enabling regional businesses to benefit from tools that better reflect local linguistic and cultural nuances.
ASEAN leapfrogs to the agentic era
ASEAN has a history of technological leapfrogging, famously skipping the PC era for mobile-first super apps connecting millions across a geographically and culturally disparate region. We may be poised to witness this phenomenon again with agentic AI in 2026, with its young, digitally savvy, and ambitious population.
Businesses unburdened by archaic legacy infrastructure have the freedom to adopt agentic AI workflows to unlock an entirely new model of work for ASEAN businesses, where AI agents and humans work alongside one another.
This would not have been possible before agentic AI. This ability to bypass old systems and traditional thinking is the essence of the leapfrog: it empowers millions of ASEAN businesses to instantly access a digital workforce of autonomous AI agents working alongside human counterparts to fundamentally transform the very nature of work itself.
– Gavin Barfield, Vice President and Chief Technology Officer, Solutions, Salesforce ASEAN
————————————————————————————————————
Heading into 2026, one of the most under-recognised shifts is the return of AI workloads back to enterprise environments. Global research shows 80% of organisations have already repatriated or are planning to repatriate AI workloads, driven by cost, compliance and performance needs. In Singapore, where tightly regulated sectors anchor the digital economy, data gravity is increasingly pulling compute closer to where data resides. The priority now is storage that is high-density, energy-efficient and built for low-latency hybrid AI — enabling organisations to keep sensitive data local while scaling AI safely and economically.
– Ban-Seng Teh, Executive Vice President and Chief Commercial Officer, Seagate
————————————————————————————————————
In 2026, APAC’s cyber landscape will be defined by systemic, AI-driven threats that outpace legacy defences, demanding a shift to proactive strategies and continuous monitoring to build resilience.
The industrialisation of crime will make attacks more frequent and sophisticated, as plug-and-play exploit kits, darknet-as-a-service offerings and AI tools make hacking accessible to everyone. Meanwhile, rapid adoption of agentic AI across IT operations will increase efficiencies but also create new high-value hacking targets. Compromised agents could autonomously disrupt services or hand over sensitive data at scale. The lines between what’s fake and real will be blurred further, as deepfakes and polymorphic, autonomous malware exploit human weaknesses and render signature-based controls obsolete.
As AI governance and regulations in APAC remain fragmented, CISOs and business leaders must invest in internal frameworks to securely manage AI agents and train internal teams against personalised scams like deepfakes. Reinforcing foundational and often overlooked network weaknesses, such as the Domain Name System (DNS) will continue to be critical. Protective DNS services, enhanced by machine learning and real-time analysis, can block malicious domains before hackers can even get close and will soon be mandatory in critical sectors.
– Lee Anstiss, Regional Director, Southeast Asia and Korea, Infoblox
————————————————————————————————————
AI integrity becomes the new trust standard
AI content authenticity and model / agent integrity will overtake data confidentiality as the foremost concern in digital trust. With 97% of APAC enterprise IT leaders having implemented or planning to implement AI agents in the next two years (according to MuleSoft’s “2024 Connectivity Benchmark Report”), organisations will demand verifiable identity and provenance for every AI asset, from training data to model outputs. Cryptographic signing, provenance tracking, and Model Context Protocol will form the backbone of new governance frameworks that authenticate, sign, and monitor models throughout their lifecycle. Boards and regulators alike will prioritise provable AI accountability, driving adoption of PKI-based standards that make authenticity and traceability the defining measures of enterprise trust.
– James Cook, Group Vice President Sales APJ at DigiCert
————————————————————————————————————
This year we’ve seen AI capture the imaginations of business leaders like no other technology before it. While AI pilot programs and experiments have generated a buzz in 2025, 2026 is the year these projects will need to generate value.
As organisations turn their attention to operationalising their initial forays into AI, many will realise that their existing processes and workflows are unsuitable to having AI bolted on top. One trend we’ll see emerge next year is the understanding that before AI can deliver value to the business, existing processes will need to first be automated and optimised.
AI is a force multiplier. If the processes AI is applied to are already broken or inefficient, those problems don’t go away – they get magnified. Ultimately, using AI and automation together is a case of 1+1=3. AI alone may create individual productivity gains, but it takes automation to make it scalable, governable, and deliver true value.
– Keith Payne, Regional Vice President APAC at Nintex
————————————————————————————————————
The verticalisation of agentic AI
In 2026, the agentic AI revolution will shift from experimentation to specialisation. Across Asia-Pacific and Japan, governments and industries will accelerate the creation of sovereign and sector-specific AI ecosystems, from healthcare and education to manufacturing and finance, each trained on contextual, localised data. This marks a new competitive era where success depends not on having the biggest AI model, but the most contextually intelligent one.
The next wave of differentiation will come from AI systems that understand local language, regulation, and nuance. As these national and industry ecosystems multiply, integration layers will play a key role in linking sovereign platforms, legacy systems, and agentic environments without compromising governance or compliance. Seamless connectivity across data, systems, and AI agents will define the next phase of digital progress in APJ. In 2026, interoperability will be the true measure of competitiveness.
The infiltration of everyday AI
In 2026, AI will no longer be a project or an experiment. It will become part of the enterprise fabric. Rather than arriving with a ‘big bang’, AI will quietly embed itself into existing workflows, managing supplier negotiations, financial operations, customer engagement, and employee assistance behind the scenes.
This is the era of agentic automation, where intelligent agents not only execute repetitive tasks but also reason, decide, and act autonomously within governed boundaries. AI systems will connect into ERP, CRM, and supply-chain environments where real work happens. The organisations that lead will balance automation speed with transparency and oversight, embedding observability and governance into every connection.
Across APJ, 2026 will mark the moment when AI stops being something separate and starts being everywhere – invisible, intelligent, and indispensable.
– David Irecki, CTO of APJ at Boomi
————————————————————————————————————
The sovereign edge will continue to evolve.
AI is a force for more distributed infrastructure as AI moves out to process data generated at the edge. Enterprises will need to consider the global management, distributed security, and remote recovery/destruction policies available for the sovereign edge and rely more on platform engineering to successfully achieve this.
As AI continues to skyrocket in adoption, businesses will look to find ways to process AI-related data locally. As a result, organisations will look to global management solutions with integrated security and edge resiliency to help keep this in check.
– Lee Caswell, Senior Vice President, Product and Solutions Marketing at Nutanix
————————————————————————————————————
A transformative year for enterprise AI
2026 is shaping up to be a significant year for enterprises across Asia-Pacific. We will see a huge investment in AI, with new frontier model builders emerging and driving substantial value creation. This trend will extend far beyond ‘ChatGPT-style’ personal assistants. retrieval-augmented generation (RAG) will flourish across the enterprise landscape, delivering outputs that are more accurate, traceable and business specific.
RAG will be especially critical for government and financial services, where data volumes are immense but need to be shielded from the open internet. It enables organisations to unlock efficiency and insight from sensitive data while maintaining data sovereignty. It will effectively operate as a mature search engine for enterprise data, moving beyond simple keyword matching to semantic understanding.
Importantly, RAG is not limited to unstructured text. It can draw from databases, images, operational systems, and a wide range of enterprise content types. Organisations are increasingly investing in building RAG-powered systems, and that trend is expected to continue in 2026.
The shift toward hyperscale AI and sovereign AI in APAC
Enterprises across APAC are increasingly adopting hyperscale AI data platforms to manage and extract value from the growing volumes of unstructured data within their organisations. These platforms integrate scalable storage, high-performance computing and built-in AI capabilities, enabling faster model development, RAG workflows and large-scale inference without relying solely on cloud ecosystems. This shift is driven by the need for organisations to maintain control over sensitive data and AI models while ensuring compliance with national regulations and governance policies.
Sovereign AI initiatives are on the rise across APAC, as organisations seek to develop AI capabilities that remain within geographic and regulatory boundaries.
In Southeast Asia, demand for sovereign and private AI approaches remains strong, especially in markets where workloads are still largely on-premises. A/NZ markets are more mature public-cloud adopters, while parts of Asia are moving toward more localised architectures.
Vietnam is a clear example: following the introduction of Decree 53, data relating to Vietnamese citizens must remain within Vietnam, accelerating the need for AI architectures that keep data in-country while supporting high-performance workloads.
Hyperscale AI data platforms help address these requirements by providing the scale, performance, and flexibility of cloud-based AI systems in architectures deployable within national borders. For many organisations across APAC, this combination of hyperscale capability and full data sovereignty is becoming a foundational requirement for the next stage of AI adoption.
– Jason Mantell, APAC Director of Solutions Architects at Cloudian
————————————————————————————————————
AI in 2026: The year data centres become the differentiator
Anyone who’s played Monopoly knows utilities are where the smart investment lies. In 2026, the most valuable square on the board is the data centre.
AI models and applications are proliferating, and the compute required to train and deploy them is prompting a rethink in how digital environments are designed, powered, and cooled. As more AI-driven services move into production, the first generation of purpose-built AI and cloud data centres will begin to come online, developed specifically for high-density, liquid- and air-cooled workloads.
This shift will have significant implications for the Asia-Pacific region. Countries such as Australia are well positioned to become hubs for AI infrastructure, supported by energy availability, institutions, undersea cable access, and a skilled workforce. These factors are expected to attract continued investment and strengthen regional credibility.
Demand for traditional cloud capacity continues to grow, while demand for purpose-built environments that can support GPU-intensive workloads, edge inference, and training on large datasets is increasing even faster. More of these workloads are expected to land in APAC hubs such as Sydney in 2026.
Just as the industrial revolution relied on factories and the dot-com boom relied on broadband, the AI era will depend on a new generation of digital infrastructure. Purpose-built AI and cloud data centres are poised to play an important role in national competitiveness.
Governments across APAC are beginning to recognise this, although policy development will need to keep pace. Countries aiming to take a leadership role in the global AI economy will require planning pathways, investment environments, and regulatory frameworks that enable timely deployment.
APAC AI spending is forecast to reach US$175 billion by 2028, highlighting the scale of the opportunity. The risk of inaction is equally clear, as delays could affect competitiveness, talent retention, and long-term industry resilience.
– David Hirst, Group Executive, Macquarie Data Centres
————————————————————————————————————
AI agents will redefine accountability in security and operations
In 2026, the rise of agentic and autonomous AI will reshape how organisations in APAC structure their security and operational teams. Instead of simply accelerating manual workflows, AI agents will increasingly take on triage, correlation, and first-response decision-making — forcing enterprises to re-examine risk ownership, auditability, and workforce design. IDC’s 2026 FutureScape for Asia-Pacific positions 2026 as the year APAC enterprises move beyond pilots to a future where AI systems act with intent, autonomy and accountability alongside humans.
Singapore is already ahead. According to Splunk’s State of Security 2025 Report, 20% of local security leaders here completely trust AI to perform mission-critical tasks — nearly double the global average. This high confidence is grounded in mature practices such as detection-as-code, adopted by 57% of Security Operations Centres (SOCs) locally, enabling teams to validate and govern AI-driven detections with discipline.
In 2026, this maturity will translate into new operational models. Organisations that embed transparency, explainability, and governance into their AI agents from the outset will be best positioned to realise the productivity and resilience gains of the agentic era.
– Robert Pizzari, Group Vice President, Asia, Splunk
————————————————————————————————————
AI investments demand measurable returns
APJ enterprises are intensifying AI investments, with over 50% reallocating funds from other areas to double down on AI, expecting AI to deliver concrete, measurable results within 12 to 18 months. APJ C-suites are also demanding a 2–4x return on investment, driving greater rigor and accountability in every step of the AI project lifecycle.
AI transforms business growth
For years, AI has been viewed as a tool for efficiency as businesses use it to automate processes, streamline operations, and reduce costs. In 2026, APJ organisations could increasingly view AI as a growth engine that drives tangible business impact and ROI.
Orchestration and trust unlock enterprise AI at scale
The companies that succeed with AI won’t be the ones with the flashiest pilots. They’ll be the ones with the best deployment strategy to orchestrate people, processes, and technology together within a well-governed and trustworthy framework.
Over 70% of APJ firms believe that orchestration will deliver a significant competitive advantage in the next 18 months. Orchestration bridges the gap between isolated workflows, allowing enterprises to coordinate, control, and optimise the work of AI agents, automated systems, and people across end-to-end workflows.
Ultimately, trust will govern AI’s growth trajectory. As APJ businesses accelerate their adoption of agentic AI, they will increasingly view data governance, explainability, security, and compliance as central to scale. Regional efforts such as the ASEAN Guide on AI Governance and Ethics highlight the push toward common principles for trustworthy AI.
Entering the era of the agentic workforce
Agentic AI will fundamentally transform how work gets done. In the future workplace, roles are redefined: AI agents think, automated systems execute structured tasks, and people lead. As AI systems become more advanced and trustworthy, humans will shift from hands-on involvement to monitoring and oversight — transitioning from “humans in the loop” to “humans on the loop.”
– Amit Khandelwal, Regional VP and Managing Director for Southeast Asia, UiPath
————————————————————————————————————
The majority of AI projects today fail not because of the technology or infrastructure, but because customers can’t trust the data feeding those models. These include issues with understanding where the data is, is it accurate, and who has access. Traditional approaches, which involve siloed tools for data security and management, don’t reflect the new world of AI and force teams into constant trade-offs between security, risk management, and business agility. We need to eliminate the challenge of managing fragmented data and enable customers to fully control and understand all their data, as well as secure it with near-zero data loss or business downtime, recover and rollback data and AI with precision.
What’s required is a single command center that helps customers understand their full data estate, while providing security, along with recovery and rollback, to unleash the value of their data for AI. This is how customers will unlock the full value of AI.
– Beni Sia, General Manager & Senior Vice President, Asia-Pacific & Japan
————————————————————————————————————
How schools will become ai’s real-world reliability test in 2026
In 2026, schools will emerge as one of Asia’s most critical real-world testing grounds for AI reliability. As adoption accelerates across the region, education systems will expose whether AI tools can perform safely, accurately, and consistently under everyday conditions, not controlled lab environments.
Classrooms operate with complexity that few other sectors experience. Students switch between languages, ask unpredictable questions, and approach tasks in different ways. Teachers are making rapid decisions much of the time with incomplete information. Learning happens across a spectrum of readiness levels. This variability presents AI with a scale of edge cases that corporate environments rarely surface. If AI can demonstrate accuracy, transparency, and safe behaviour in schools, it is likely robust enough for wider enterprise use.
This shift has significant implications for Southeast Asia’s digital future.
First, digital citizenship will move from a curriculum add-on to a regional priority.
Southeast Asia’s digital economy continues to expand. The World Economic Forum projects that increased digital participation in Southeast Asia has the potential to drive the region’s digital economy to reach up to $1 trillion GMV by 2030.
Digital economic growth happens as a result of digital inclusion and the active participation of digital users. As AI becomes embedded in workplaces, students must develop the ability to question outputs, recognise bias, and understand when to rely on AI, and when not to. These instincts form early, often during a student’s first interactions with generative tools. Schools will set the baseline for how the next generation interprets, evaluates, and collaborates with intelligent systems. The quality of these early experiences will shape the region’s long-term digital capability.
Second, edutech developers will need to shift from automation-first design to pedagogy-first design.
Schools are not seeking tools that shortcut learning. They are looking for systems that support “productive struggle”, the cognitive effort required to build understanding. In 2026, successful AI-enabled education products will guide students through reasoning, prompt reflection, and surface misconceptions, rather than simply generating answers. For developers, this means designing AI that slows thinking down when needed, models curiosity, and supports conceptual growth. Tools that compress learning into efficiency exercises will lose relevance.
Finally, responsible AI adoption in education will become a reputational and operational benchmark.
Governments and school networks are accelerating guidelines for transparency, data protection, and age-appropriate use. As a result, vendors entering the education sector will face stronger expectations for explainability, safety controls, and alignment to learning outcomes. Solutions that cannot withstand the scrutiny of school environments will struggle to scale in other regulated industries.
In 2026, the question for AI will shift from “What can it do?” to “Can it be trusted every day, with every learner?” Education will be where that trust is earned – or lost. For Southeast Asia, this makes schools not just adopters of AI, but key architects of its future reliability.
– Tanya Hawkes, Director of Teaching & Learning, Stamford American International School
————————————————————————————————————
Voice AI will become the new interface layer
Voice AI is entering a new phase in Asia-Pacific, shaped by the rise in enterprise AI investment, the region’s linguistic and cultural diversity, and increasing expectations for real-time, empathetic customer service. While earlier efforts focused on text-based chat and automation pilots, enterprises are now enabling systems that can manage live dialogue, retain context, interpret sentiment, and switch languages seamlessly. In 2026, these capabilities will move from experimentation to the default standard for enterprise-grade conversational systems.
– Amitabh Sarkar, Vice President & Head of Asia-Pacific and Japan – Enterprise at Tata Communications
————————————————————————————————————
Intelligent global business services will evolve into the strategic growth engine of 2026
Decades of siloed SaaS tools have left employees juggling apps that don’t talk to each other, slowing decisions and creating frustration. Even in Singapore, where digital ambition is high, AI is still being built in fragments: ServiceNow’s Enterprise AI Maturity Study shows only 26% of leaders feel their organisation is mature enough to transform with AI, while 72% deploy AI through multiple internal task forces.Expectations are shifting: complex systems should feel simple. Traditional GBS models built on labour-intensive workflows are rapidly becoming obsolete. Artificial Intelligence has emerged as the core infrastructure for service delivery, powering predictive operations, self-service experiences, and self-healing processes that eliminate inefficiencies before they occur. This marks a fundamental shift from cost-centric thinking to value-driven strategies, positioning GBS as a strategic growth engine rather than a back-office function. Organisations are responding decisively: technology budgets for GBS are rising, and AI is now viewed as indispensable, not only for achieving cost leadership but for unlocking innovation, resilience, and enterprise-wide value creation.
Empathy by Design Will Separate Successful AI Deployments from Failed Ones
As AI becomes more autonomous in 2026, the companies that succeed with agentic AI will be those that invested in understanding human-AI interaction patterns, that user-tested relentlessly, and that prioritised making people feel understood over showcasing technical prowess. The future isn’t just autonomous, it’s a tight partnership between Human and AI. For example, web agents that learn from human actions to complete tasks across applications and websites without APIs or integrations. This evolution reflects broader APAC trends: IDC predicts that by 2028, consumers in APAC will spend $32 billion through AI agents, highlighting their potential to drive value.
Data continues to be the living ingredient for reliable AI
By 2026, enterprises will realise the true bottleneck in AI adoption isn’t model capability but data quality. ServiceNow’s Enterprise AI Maturity Study shows Singaporean leaders most often cite data security (21%) and an AI governance deficit (15%) as their top barriers to getting value from AI.
Just as a great soup depends on clean, well-prepared ingredients, governance frameworks, and a skilled chef, AI delivers seamless workflows only when high-quality, unified, and well-governed data is combined with clear standards and human oversight. Organisations that get this right will turn fragmented initiatives into operational clarity, enabling accurate predictions, fair decisions, and experiences that feel effortless for employees and customers alike.
Hands-on talent will drive trusted AI in regulated industries
As AI scales across complex, regulated environments, the demand for a workforce skilled in both AI tools and responsible governance will intensify. Hands-on, practical learning will become central to closing this gap. Enterprises will increasingly invest in programmes that let employees and students build real-world AI solutions in secure, compliant environments, preparing talent to deploy AI responsibly. Initiatives in regulated industries, such as the Monetary Authority Singapore’s Pathfinder Programme, demonstrate the value of structured, collaborative approaches to AI development and deployment.
– CK Tan, APJ Innovation Officer, Singapore at ServiceNow
————————————————————————————————————
Demand for context engineering will grow as Agentic AI booms
The growth and reliability of agentic AI will hinge on accurate context engineering, which ensures AI systems access and utilise the right data at the right time. In 2026, context engineering will become critical as enterprises struggle with scattered data across unstructured sources like documents, emails, apps, and customer feedback.
Effective agentic AI requires relevant data inputs to deliver accurate responses. Many failures in AI development trace back to the inability to provide relevant context for applications. Context engineering addresses this challenge by facilitating precise data retrieval, governance, and orchestration, enabling agents to seamlessly identify, retrieve, and process owned data.
There are a limited number of platforms offering comprehensive context engineering capabilities at this time. Demand for such solutions will rise sharply in the next year. Businesses will increasingly seek AI platforms that integrate context engineering at their core, boosting the adoption of contextually aware and reliable AI systems.
– Ken Exner, Chief Product Officer at Elastic
————————————————————————————————————
Data infrastructure will become southeast asia’s AI differentiator in 2026
In 2026, the ‘AI advantage’ will evaporate for companies that rely on generic models alone. As AI tools become easier to build and increasingly interchangeable, the real advantage shifts to the quality, connectedness, and trustworthiness of the data behind them. Early adopters already prove this: 92% are seeing ROI from AI, yet many still struggle with fragmented systems and data that isn’t ready for machine learning. The biggest gains now belong to the companies that fix the foundation, not the ones chasing the flashiest model.
The winners are those that master the “data flywheel” — unique data fuels AI, smarter AI produces even more unique data. This cycle builds a lasting competitive edge. In a region as dynamic and digitally-driven as Southeast Asia, this shift will redraw competitive lines fast.
In a world where AI tools are becoming commodities, the real advantage goes to enterprises that prioritise data quality, accessibility and identifying use cases with clear business impact.
– Satchit Jogeklar, Managing Director, ASEAN, Snowflake
————————————————————————————————————
AI governance takes centre stage as governments grapple with AI regulations and digital sovereignty
Governments around the world will increasingly pursue “Sovereign AI” solutions to ensure control over data and compute resources within their borders, driving the creation of national AI ecosystems and regional data centres.
– Vrushali Sawant, Data Scientist, SAS Data Ethics Practice
A bumpy ride of AI-driven workforce transformation will require a new approach to skill development
Public agencies will capture institutional knowledge by training retrieval augmented generation (RAG) systems on the documented expertise of senior staff, creating AI mentors that provide junior employees with on-demand access to decades of accumulated wisdom and best practices. Unfortunately, this will also lead to more instances of AI sabotage where workers, concerned with being replaced by AI, purposefully produce poor content to contaminate the training.
– Steven Tiell, Global Head, AI Governance Advisory, SAS
AI will be both enemy and ally of public sector investigators and tax officials
With more sophisticated generated identities and transactions (given widespread use of generative AI platforms by fraud rings), and ever-more sophisticated tax avoidance schemes, agencies will need to place an even greater emphasis on fraud detection, identity verification and tax-related financial data analysis to lower the risk and safeguard tax revenues.
– John Bace, Industry Consultant, Tax & Revenue Compliance, SAS
In 2026, we will see identity management become the backbone of inter-agency agreements to ensure lawful, fair, and secure data exchange. This will promote cross-agency data sharing, enabling more contextual fraud, waste and abuse risk detection and mitigation.
– John Stultz, Principal Solutions Architect, Risk, Fraud & Compliance, SAS
Citizens and revenue agencies will benefit from the spread of real-time analysis in tax agencies. This will help reduce account takeover threats and provide in-the-moment feedback to improve filing accuracy and reduce tax gaps.
– Carl Hammersburg, Senior Manager, Government and Healthcare Risk & Fraud, SAS
————————————————————————————————————
AI transparency will become an enforceable consumer right
The AI identity crisis is here. Our Twilio research shows 90% of consumers fail to correctly identify AI-generated voice clips. When customers cannot distinguish who they are speaking to, regulatory intervention is a certainty.
In 2026, transparency will transform from an ethical talking point to an enforceable consumer right. Mandatory and explicit disclosure – example, “I am Ruby, [Spokesperson]’s Voice AI assistant” – will become a core compliance requirement across finance, retail, telecom, and government. Brands that proactively build transparent AI interaction frameworks will gain a competitive edge.
The AI trust threshold will be the new CX frontier
In 2026, trust-aware routing and AI observability will be non-negotiables in a mature CX environment. Brands will use trust as a new routing rule, determining when AI should lead the conversation, and when a human takes over.
Savvy brands will monitor for the signals of sentiment decay, customer hesitation, or rising uncertainty in real time. Thanks to advanced sentiment analysis and observability tools, companies can now route conversations based on these trust signals, not just issue categories. This prevents AI missteps from escalating into significant churn.
– Nicholas Kontopoulos, Vice President of Marketing, Asia Pacific & Japan, Twilio
The future of conversational AI will be modular and flexible
The high cost and poor performance of rigid, all-in-one platforms will drive a dramatic architectural shift in conversational AI across APJ. With 81% of APJ directors finding it costly to keep up with rapidly evolving models, brands can no longer maintain monolithic systems that age quickly.
Performance failures at the local level compound this urgency: 40% of APAC consumers found that AI agents failed to understand their accents over the phone. To mitigate cost and performance pressures, brands will move towards adopting modular, plug-and-play systems. This approach supports Bring-Your-Own-LLM, which allows flexibility to swap out models cost-effectively, avoid vendor lock-in, and integrate specialised AI tailored for the region’s nuanced language expectations.
– Christopher Connolly, Director of Solutions Engineering, Twilio
————————————————————————————————————
In 2026, AI governance in Asia will become a more regular part of board oversight. Many organisations have introduced AI across operations to improve efficiency, and some are beginning to explore agentic systems that act autonomously in workflows. As use cases expand, boards will need clearer visibility into how AI behaves over time and how outcomes are reviewed and escalated when necessary.
Diligent’s 2026 Governance Outlook Report shows that while more than half of surveyed organisations in Asia have already adopted AI in specific areas of their operations, most boards have not yet formalised training or governance processes to support oversight. This reflects a gap between deployment and supervision. In the year ahead, customers, partners and regulators are likely to ask for clearer evidence of how AI systems are monitored day-to-day – similar to how cyber and third-party risk oversight became routine in recent years.
To provide the necessary level of assurance, we anticipate that boards will increasingly rely on the teams who are responsible for AI implementation, such as IT teams, CIOs, and CTOs. For example, for IT decision makers, AI operational reporting will become more strategic, focusing not only on technical performance but on the information boards need for governance – where AI is used, how decisions are made, and how outcomes are reviewed. The same report shows that 37% of organisations in Asia already include CIO or CTO participation in board discussions on AI, and this involvement is expected to grow as AI becomes more embedded in workflows. Their role is to provide a clear view of where AI is used, how systems perform in real scenarios, and when human intervention is required to maintain accountability.
– Dusk Lim, Principal of Enterprise Growth APAC, Diligent
————————————————————————————————————
AI is in its early days, and the pull-ahead will be faster
AI continues to evolve at a pace where we have not seen clear limits. It reminds me of the early days of the internet, when most people only used it to read news. The most significant shifts in AI are still ahead of us.
In this environment, speed matters more than anything else. We are entering a phase where companies that move quickly on AI adoption will gain a competitive edge over peers that are slower to adapt, regardless of size or history. It is very much a “fast fish eats slow fish” era. Teams that learn quickly from real-world AI usage will widen their lead, while slower organisations will find it increasingly difficult to catch up.
AI is also shifting from primarily entertainment-focused applications to a mix of entertainment and real business value. This transition will open many opportunities for companies that can move at the pace the technology requires.
A striking trend is the rise of young, small AI companies that move very quickly. While many did not exist a few years ago, their growth and compute consumption already exceeded those of traditional IT or cloud companies. Their survival depends on constant iteration — which again reinforces the “fast fish” dynamic.
Enterprise AI will hit a real breakthrough moment
The biggest opportunities will eventually open up in the enterprise space. E-commerce, advertising, and media are already showing strong adoption, particularly in consumer-facing applications. In e-commerce, AI-powered search is reshaping how consumers discover and buy products. Generative AI is also enabling hyper-personalised campaigns at scale. Users can now upload a photo and receive high-quality, personalised images or video creatives within seconds, transforming how brands engage with consumers.
The next wave will come from education, gaming, and financial services. This shift isn’t limited to start-ups or AI-native companies; traditional industries are also undergoing significant transformation as they adopt AI across operations.
The financial services industry, for example, is increasingly adopting AI to build internal knowledge bases for employees to swiftly retrieve precise, context-aware information, which is critical for accurate decision-making. At the same time, AI-powered agents are also transforming customer services by delivering personalised support.
– Yongliang Zhang, General Manager, BytePlus
————————————————————————————————————
2025 has been a decisive year in Southeast Asia’s AI journey. Organisations have moved beyond early experimentation to deploying agentic AI systems that can act, decide, and drive outcomes at scale. The biggest lesson is clear: AI delivers the most value when humans stay in the lead.
Accenture’s research reinforces this shift. The companies realising the strongest gains in productivity, accuracy, and innovation are those that put human judgment at the centre, with AI scaling execution around it. In these high-performing organisations, people and AI continuously learn from each other — but it is human expertise, intent, and ethical grounding that set direction, refine outputs, and define what “good” looks like. This human-in-the-lead approach has proven essential for navigating Southeast Asia’s complex markets, diverse consumers, and fast-moving digital economies.
Looking to 2026, the principle of humans in the lead, AI at scale will define how organisations scale AI responsibly and effectively. Humans make the critical decisions while AI handles volume, variability, and speed — freeing people to focus on judgment, creativity, and solving higher-order problems. The organisations that will thrive are those that combine human judgment with AI-driven execution, embedding intelligence across operations while keeping people at the decision-making helm. When humans define the purpose and AI amplifies execution, businesses unlock unprecedented value and accelerate reinvention at a scale neither could achieve alone.
– Anoop Sagoo, CEO, Southeast Asia, Accenture
————————————————————————————————————
Perhaps the most exciting shift for 2026 is how AI changes who can participate in innovation. Natural‑language interfaces and agentic AI allow domain experts, doctors, plant managers, supply‑chain leaders, to orchestrate AI workflows without being specialists. Combined with secure, well‑governed infrastructure, this democratizes innovation and accelerates time to value. At the same time, sustainability has become a prerequisite for continued innovation: Advanced cooling and efficient systems are essential to deliver more AI performance per watt, while reducing energy use and emissions. The leaders of 2026 will be those who embed responsibility, sustainability, and
inclusivity into their AI roadmaps, always putting people at the center.
– Sumir Bhatia, President, Asia Pacific, Infrastructure Solutions Group, Lenovo
————————————————————————————————————
The Biggest AI Shifts Enterprises Should Prepare for in 2026
AI adoption will enter a new phase in 2026, as enterprises shift from hype-driven experimentation to the hard engineering work required to deploy AI at scale. Kyndryl’s 2025 Readiness Report highlights the magnitude of this shift: 54% of organisations now report positive ROI from AI, yet 62% of projects remain stuck in the pilot phase, and only 29% of leaders feel prepared to manage future AI risks. This readiness gap will define the year ahead. Scaling AI now depends less on new models and more on trusted data pipelines, resilient infrastructure, governance frameworks, and the ability to integrate AI into mission-critical systems.
Businesses will also make more deliberate moves toward agentic AI, introducing agents into workflows across IT operations, finance, supply chain, and customer service. But as adoption grows, so will expectations around governance and accountability. Structured approaches will help organisations determine which tasks can be automated, where human judgment must remain essential, and which hybrid roles must be created for safe and responsible deployment. In 2026, AI transformation becomes as much about organisation and workforce design as it is about algorithms.
As AI becomes deeply embedded in enterprise infrastructure, organisations will need to secure models, data pipelines, training datasets, orchestration systems and inference environments, not just traditional networks. Boards will treat AI security as a strategic priority.
Ultimately, 2026 will be the year enterprises recognise that AI ROI is inseparable from modernisation. Those that invest in resilient infrastructure, robust data foundations and secure-by-design AI operations will move from pilots to production — and turn AI ambition into measurable business impact.
– Andrew Lim, Managing Director, Kyndryl ASEAN & Korea
————————————————————————————————————
By 2026, AI-powered agents will evolve from managing routine tasks to automating complex processes and operating autonomously with minimal human oversight. These systems, capable of understanding intent and anticipating user needs, will transform organisational operations.
In APAC, rapid AI adoption is driven by structural challenges, labor shortages, cost pressures, and rising customer expectations. Multi-agent systems are already delivering up to 50% efficiency gains in sectors like finance, IT, and customer service. Research by Zebra Technologies and Oxford Economics reveals that top companies in retail, manufacturing, and logistics could unlock an average of US$3 billion in additional revenue and US$120 million in profit by improving frontline workflows with AI, highlighting the transformative potential of AI-led automation. Examples like OpenAI’s collaboration with Spotify demonstrate how AI enhances discovery, personalization, and conversational interactions. As APAC advances in AI maturity, industries should explore similar innovations to boost productivity and user experience.
Embedding AI agents into devices enables autonomous inventory tracking, predictive maintenance, and task optimization. AI orchestration tools can further enhance efficiency, minimise downtime, and empower frontline workers to focus on higher-value activities.
– Christanto Suryadarma, Vice President for Southeast Asia (SEA), South Korea and Channel APJeC, Zebra Technologies
————————————————————————————————————
AI Governance Will Become the New “DevOps” for the Enterprise
In 2026, enterprises will realize that AI adoption isn’t primarily a modeling challenge — it’s an operational one. As every team from development to security to business operations begins using AI tools and agents, CIOs and platform leaders will be forced to standardize how AI is discovered, approved, secured, and monitored across the company. The result: AI governance will evolve into a new enterprise discipline, much like DevOps did a decade ago. Companies that treat AI as a governed supply chain, rather than a collection of disconnected pilots, will scale faster and avoid compliance and security pitfalls that slow their competitors.
Just as DevOps faded into the background so developers could focus on building, AI governance will mature to the point where developers barely realize it’s there.
Shadow AI Will Become a Bigger Risk Than Shadow IT
With open-source models, SaaS AI tools, and API-based agents now one click away, organizations will see a surge in “shadow AI” — teams adopting unvetted models outside formal processes. In 2026, this will eclipse shadow IT as the top operational risk CIOs face. Security teams won’t just worry about rogue infrastructure; they’ll worry about unapproved models with hidden vulnerabilities, poisoned datasets, and undocumented behaviors. To counter this, enterprises will adopt centralized AI catalogs and enforce model allow-lists as standard practice, similar to how software artifact governance became mandatory during the DevOps era.
Companies Will Shift Standalone Models to Deeply Integrated, Context-Enriched Systems
While today’s AI adoption often starts with generic LLMs and isolated prototypes, enterprises are realizing that real value doesn’t come from the model alone — it comes from how well that model is connected to their internal systems. In 2026, the focus will move away from “building your own” models and toward deploying AI that natively integrates with internal assets: data sources, tools, APIs, operational workflows, and governance layers.
Models and agents will increasingly use MCP-like connectors to enrich prompts with internal organizational context, retrieve real-time business data, and perform actions across existing enterprise systems. This shift turns AI from a static text generator into an operational participant — one that queries, validates, updates, and orchestrates tasks based on live internal information.
As a result, companies will reduce drift, improve reliability, and unlock far faster time-to-value. Instead of experimenting in isolation, enterprises will rely on integrated, governed, production-ready AI systems that understand their business, operate within their environment, and continuously stay aligned with their internal truth.
– Yuval Fernbach, VP & CTO of MLOPs, JFrog
————————————————————————————————————
2026 will be the year where ROI will be redefined for AI adoption. We have gone past the “pilot and prove” stage in 2025, as APAC’s leaders now know what they want to build, and are building it. Confluent’s 2025 Data Streaming report showed that 56% of APAC businesses have already deployed chatbots, copilots, and AI assistants, outpacing Europe and North America.
While AI adoption has surged, so has scrutiny. ROI pressure has moved from the boardroom to every business unit, and many executives are now prioritizing measurable payback. Cost-control will remain top-of-mind, and 2026 will be the year boardrooms define ROI as not just amounts saved, but also growth enabled, market share captured, and possibilities unlocked. Success with AI will be increasingly defined by its impact on the company’s ability to grow and thrive, rather than survive.
As ROI definitions shift, boards will double down on initiatives that master continuous intelligence and quietly defund projects that don’t. How does investment in anti-fraud agents contribute to greater customer confidence? How will customer relationship management AI models drive greater spend? Answering these questions will help AI project owners demonstrate how each investment moves the needle. In 2026, I expect fraud, security, and customer operations to remain the clearest ROI lanes for AI. AI project owners must prioritize presenting clear benefits and financial justifications up front.
– Nick Dearden, Field Chief Technology Officer, Confluent
————————————————————————————————————
Commercial teams will rely on AI foresight
More commercial and route-to-market teams are set to increasingly adopt AI, shifting its role from a mere reporting tool to a predictive engine that enables real time insights into customers, consumers and overall market dynamics. Companies are already leveraging machine-learning models, trained on years of granular store-level data, to streamline operations, optimize routes, and recommend product assortments for micro-markets. The bigger leap will be AI that anticipates demand at the specific street and individual store level, allowing for unparalleled precision in planning and execution.
Factories will become learning systems, not just automated ones
Manufacturing will shift as factories evolve beyond automations into true learning systems. We are already seeing early examples like computer vision detecting micro-defects invisible to the human eye, 3D printing reducing reliance on spare parts inventory, and VR environments accelerating onboarding for factory talent. By 2026, more plants will adopt “predictive automation” – systems that use continuous data collection and analysis to self-optimize energy usage, maintenance cycles, and quality checkpoints autonomously. The real transformation won’t come from replacing workers, but from using these learning systems to augment their judgment with continuous, data-driven insights.
Supply chains will prioritize resilience over efficiency
The past few years have proven to FMCG companies that efficiency alone is not a sustainable strategy. In 2026, AI will play a central role in balancing necessary agility with resilience. Technology will enable the sensing of disruptions early, allowing companies to adjust before issues cascade into crises. This involves using AI to predict upstream raw material risks, dynamically re-route logistics, and maintain robust service levels in volatile markets where consumer behavior can shift rapidly. The goal is no longer just lean operations, but intelligent, adaptive supply chains that can weather any storm.
Technology that strengthens the frontline
The FMCG industry will always be fundamentally people-driven, powered by individuals ranging from farmers and factory teams to frontline sales representatives. Technology’s role should not be to replace this workforce, but to elevate it. The objective is to eliminate repetitive tasks from their day, allowing them to focus on the high-value aspects of their jobs that truly move the needle for the business. The companies that stand out in 2026 will be those using technology to speed up everyday decisions and clear operational bottlenecks, empowering the teams who make the business run.
– Jeal Desai, Director of Mondelez Digital Services, SEA Mondelez International
————————————————————————————————————
AI becomes practical – and fit-for-purpose models will take center stage
Business leaders will need to rethink their infrastructure strategies to support more diverse and demanding AI workloads. We will see growing interest in unified inference layers that can support a wide range of AI models without compromising performance and cost efficiency. At the same time, there is strong momentum around connecting enterprise application platforms with cloud-based AI accelerators, giving organizations a more seamless way to operationalize AI at scale. By pairing flexible platforms with specialised computing, enterprises can accelerate the shift from pilots to producing measurable business impact.
Virtualisation evolves to meet the demands of AI-era workloads
AI is reshaping how enterprises think about infrastructure. Traditional virtualization approaches, built for predictable and uniform workloads, are now being stretched by the needs of modern AI — which demand higher performance, lower latency, and far more flexibility.
In 2026, enterprises will increasingly adopt virtualization strategies that bring together virtual machines, containers, and specialized compute under a single operational model. This helps platform teams modernize at their own pace while supporting both existing applications and new AI-driven workloads.
Hybrid cloud becomes the default architecture for modern AI
As AI models increasingly rely on real-time data, distributed systems, and specialized computing resources, enterprises need architectures that allow them to run workloads as close to their data as possible, while still maintaining scalability and resilience.
The demands of AI require the hybrid cloud. And in 2026, hybrid cloud will solidify its position as the standard operating model for intelligent enterprise systems. Organizations will prioritize platforms that help them maintain control over sensitive workloads on-premises, scale using public cloud capabilities, and bring intelligence closer to where data is generated at the edge.
Governance frameworks reshape digital strategy across APAC
In 2026, enterprises will increasingly prioritise AI systems that can be audited, monitored, and governed across hybrid environments, ensuring that decisions remain traceable and models behave as expected. This governance shift will also influence architectural choices, vendor selection, and skill priorities. Enterprises will seek open, trustworthy solutions that allow them to examine how models are built, how data is used, and how decisions are made. In regulated industries like financial services, these capabilities will become non-negotiable.
Skills, communities, and collaboration become the real accelerators
The demand for cloud-native, AI, and cybersecurity talent continues to outpace supply across APAC, and in 2026, the gap will only widen unless organizations invest in a skills-first approach to build, operate, and optimize modern digital systems.
Open source communities will play a central role in this shift. They provide shared knowledge, transparency, and a global ecosystem rooted in collaboration. Tools and frameworks are also made available to everyone, instead of just a few. As more enterprises contribute back to these communities – by building on ideas quickly and responsibly – APAC will strengthen its position in digital innovation, not just as a consumer but increasingly as a creator.
– Guna Chellappan, General Manager for Singapore, Red Hat
————————————————————————————————————
2026 will be the year APAC enterprises shift from experimenting with AI to operationalising it at scale. From my conversations with leaders across the region, it’s clear they are no longer asking how to pilot generative or agentic AI. They’re asking me how to embed it into decision cycles, workflows, and customer-facing processes with confidence. This shift will reshape data architectures, governance models, and the skills enterprises prioritise in the year ahead.
Demand for trusted data rises as AI goes operational
Enterprises will judge AI not by prototypes but by measurable business outcomes in the new year. Agentic AI systems capable of planning, executing, and adapting tasks will become more common in operations, risk management, and supply chain optimisation. But adoption will only accelerate if organisations close the trust gap.
I see APAC enterprises increasingly demanding explainability, transparent lineage, and governed data to ensure these autonomous systems act reliably. Black-box models will be sidelined in favour of architectures that support traceability and compliance. This makes open, interoperable data foundations not just desirable but essential.
Open Lakehouse architectures become the backbone of AI execution
As multi-cloud becomes the norm, tightly coupled data stacks will struggle to keep pace with the demands of AI. I believe the Open Lakehouse will solidify as the preferred foundation for enterprises seeking flexibility without sacrificing governance or performance in 2026.
Open table formats, especially Apache Iceberg, will become the de facto standard for AI-ready data. Iceberg’s ability to decouple storage from compute, maintain transactional consistency, and support rich metadata gives enterprises the portability and transparency they need to run AI across diverse environments.
Interoperability to become a strategic requirement
With AI embedded deeper into processes, enterprises will prioritise systems that seamlessly connect across clouds, applications, and operating environments. I expect the next wave of innovation to come from interoperable AI pipelines, where data ingestion, transformation, analytics, and action flow as one continuous loop.
Enterprises will start looking beyond dashboards to put action directly in the flow of work, triggering automated decisions, alerts, and agentic workflows within business applications.
– Maurizio Garavello, SVP for Asia Pacific & Japan, Qlik
————————————————————————————————————
Agentic AI is widely considered the next frontier in cybersecurity for its ability to adapt, learn, and execute actions on its own, absent of any human input or intervention. Despite its promise, the industry is far from seeing a fully autonomous Security Operations Centre (SOC).
Instead, in 2026, we push for the transformation of the SOC, not by automating how SOCs work, but by reinventing how SOCs work with an expert-based approach. We will see the growing use of agentic AI taking the lead with human aid, versus human leading with AI aid. This will radically transform how SOCs work by augmenting human expertise with advanced automation and AI, without completely replacing the need for skilled analysts. With large language model (LLM) adoption already high in Singapore, with 77% of local IT leaders acknowledging themselves as repeat users, human oversight and processes remain critical, even with AI enhancing SOC efficiency.
– Dan Schiappa, President, Technology & Services, Arctic Wolf
————————————————————————————————————
It’s judgement day for the AI bubble
95% of Generative AI investments have produced zero results. The industry is bracing for a reality check as AI transitions from experimental pilots to robust AI applications that have the rigor to stand up to everyday use in industries that demand real-time delivery.
Data secured by architecture that serves up the right data – to avoid intermingling and prompt injection
The risk of prompt injection and data intermingling is high. The solution is architectural: using an Agent Mesh to ensure AI agents are fed only the necessary, filtered data, not raw, sensitive documents.
Forget prompt engineering – helping AI to respond better with context engineering
Forget basic prompt engineering. The future is Context Engineering: building architectures to swap memory and rules of engagement so AI can operate and react with up-to-the-second awareness for decision-making.
Meet the AI ‘team’ – the rise of multi-agent systems
A single agent cannot be an expert in everything. 2026 will see the rise of orchestrated AI “teams” (multi-agent systems) communicating via A2A protocols to solve complex enterprise workflows securely and efficiently.
In the year ahead, AI’s competitive edge will come not from just better models or smarter prompts, but from connecting AI to the live, operational pulse of the business from day one.
– Edward Funnekotter, Chief AI Officer, Solace
————————————————————————————————————
2026 will be the year when agentic AI evolves from a good-to-have to a mission-critical technology that will power core processes across the organisation. This extends from customer experience (CX) to how employees collaborate, learn, and stay connected.
From a CX standpoint, AI’s true value goes well beyond managing routine queries or automating repetitive work. It will also be able to determine when a virtual agent should engage, which type to leverage (scripted bot, agentic model, or voice assistant), and when to
hand off to a human. Hand-offs could be based on factors like cost, impact, and experience to determine the optimal human-AI balance in each and every interaction. This is particularly important in this region. Zoom-commissioned research on AI natives in APAC – workers aged
18 to 24 who have grown up with AI – found that they want speed and efficiency from the technology, but also the option to escalate to human support when complexity or nuance matters. As organisations come to establish their own models of human–AI collaboration, we
will see these virtual agents becoming key members of the team who can be measured against performance, cost, and experience metrics – as with any human employees.
Employees themselves will spend more time on meaningful tasks thanks to agentic AI embedded into everyday workflows. We already know that intelligent agents are capable of taking on manual and repetitive tasks, whether that’s updating project statuses, scheduling meetings, summarising discussions, and managing follow-ups. But in 2026, we could see agentic AI taking that to the next level with its ability to quickly move conversations into completion. Think recommendations on meetings to skip based on prior knowledge about the user’s role and participation in the meeting, or proactive prompts in advance of a meeting so users know the agenda, previous action items, and key insights. Employees can then focus their time and energy on delivering the creativity, strategy, and human connection that remains critical in the age of AI.
The key challenge, however, is that AI quality has become a significant point of friction for organisations looking to extend their agentic AI capabilities. The same research found that one of the biggest frustrations of using AI at work amongst APAC’s AI natives is the need to
manually correct AI outputs. We anticipate that to solve the problem of quality, more organisations will embrace a federated AI approach, turning to multiple models to achieve higher accuracy, flexibility, and cost efficiency. This way, organisations can ensure their AI systems remain adaptable, resilient, and future-ready as they scale agentic AI across their business.
– Steve Rafferty, Head of APAC and EMEA, Zoom
————————————————————————————————————
In Asia Pacific (APAC), fewer than 1% of organisations have successfully scaled AI across their business. In 2026, the priority will shift from experimentation to purposeful integration. This needs to start with selecting tools that directly support each team’s objectives while avoiding a fragmented tech stack. Unified platforms that integrate popular apps will become essential to bring together workflows, make knowledge discoverable, and keep information organised.
Another critical shift will be bridging the gap between personal and organisational AI use. Much like the early cloud era, adoption is spreading bottom-up – employees are already using AI for drafting, summarising, and automating routine tasks. Organisations that harness this grassroots momentum into secure, structured workflows will unlock real, measurable impact. Context-aware AI will be key here, interpreting not just tasks but the broader work environment — shared archives, team goals, and ongoing projects — so systems can act with organisational awareness rather than just task-level intuition.
Despite the increasing reliance on automation, the human element will remain central. Companies will need to cultivate an AI-confident workforce by investing in skills and fostering a safe, collaborative environment where employees can experiment, share insights, and learn continuously. AI won’t replace judgment or creativity, but teams who know how to collaborate effectively with it will drive the next wave of productivity.
Finally, how we measure productivity will evolve. Output alone will no longer suffice; impact will be measured by how quickly insights are generated, decisions improved, and work moves from idea to execution. If 2025 was the year of experimentation, 2026 will be the year AI becomes an integral, dependable part of everyday work.
– Matthew Hong, Asia GTM Lead at Dropbox
————————————————————————————————————
Recovery and resilience: Addressing the dual challenge of AI-driven attacks and expanded digital surfaces
AI significantly accelerates the pace of attacks and expands the attack surface that malicious actors leverage. There is an increased urgency for CISOs to adopt an “assume breach” mindset and prioritize ensuring data integrity and recovery. When an attack occurs, the time to get a business up and running is the critical metric. However, in 2026, the new imperative is to ensure data integrity and the ability to recover to a verified, clean point quickly. AI tools can rapidly generate malware and exploit known vulnerabilities. Organizations must pivot to recovery strategies that utilize integrity validation and isolated “cyber vaults.” The recovery strategies will guarantee the restored environment is free of malicious code, making robust recovery engines a necessity, not a convenience.
Identity security: Identity-based attacks will dominate CISO investments
The scale of non-human identities in the AI era will become a critical vulnerability. Attackers continue exploiting the labyrinth of non-human credentials; however, in 2026, they’ll achieve full-system compromise. A recent survey revealed that 89% of organisations plan to hire professionals in the next 12 months specifically to manage identity security. Identity infrastructure will become more critical than the data infrastructure it protects.
The great AI sprawl
The proliferation of AI agents is creating the “great AI sprawl,” forcing IT and security teams to reconcile rapid deployment with system control. The dynamic will necessitate a governance renaissance in 2026 and immediate, focused investment to bring agents into production safely and at scale. To achieve production-grade agent deployment, organizations must rapidly implement monitoring and governance controls to ensure visibility into which applications or data agents are accessing and that they adhere to corporate policies. Inevitably, agents will make mistakes, and they will need to have remediation strategies in place. Organisations will need to overhaul their current IT and security workforce management. In 2026, heavy investment in robust security and governance systems will be essential to monitor, control, and remediate agent output.
The convergence mandate: Multi-cloud chaos forces unified control plane or enterprise extinction
In 2026, the myth that native cloud tools are sufficient collapses as organizations recognise their siloed multi-cloud environments are severely slowing down cyber recovery. Using multiple native backup tools leads to long restoration times and frequent emergency
migrations. Recovery speed will become the only metric that matters as a unified multi-cloud platform transforms from a convenience feature to a non-negotiable survival requirement. The most resilient organizations will consolidate control under a unified plane, recognizing
that identity is the central hub for their entire multi-cloud data environment, demanding the seamless integration of identity security with data protection.
– Arvind Nithrakashyap, Co-Founder and Chief Technology Officer, Rubrik
————————————————————————————————————
AI agents will move from experiments to autonomous enterprise workflows
In 2026, AI agents will evolve from proofs-of-concept to fully embedded digital co-workers. These agents will handle multi-step tasks – summarising information, coordinating workflows, preparing reports, and triaging support needs – reducing routine workload for teams across departments. As autonomy increases, enterprises will invest heavily in oversight: defining guardrails, ensuring transparency, and maintaining human accountability. Organisations will shift from asking whether agents can perform tasks to how responsibly, safely, and consistently they can do so.
Human outcomes become a primary metric for evaluating technology investments
Technology decisions will increasingly be evaluated through a human lens. Traditional metrics like throughput and uptime will remain relevant, but organisations will also measure technology based on its ability to reduce digital fatigue, minimise friction, and support higher-value work. As AI takes on more administrative and repetitive tasks, the focus will shift toward empowering people – creating technology experiences that improve clarity, creativity, and fulfilment, while also strengthening engagement and long-term performance.
– Jennifer Baile, Vice President, Global Services & Solutions, Greater Asia, HP Inc
