수십 년 동안 IT 운영 매뉴얼은 대개 50페이지 분량의 빽빽한 PDF 문서였다. 사람이 만들고 사람이 읽도록 설계된 문서는, 감사가 필요해질 때까지 디지털 저장소 어딘가에서 방치되는 경우가 대부분이었다. 그러나 2026년에 접어든 지금, 전통적인 SOP는 사실상 수명을 다한 상태다. 이제 이 매뉴얼의 주된 사용자가 사람이 아니기 때문이다.

시스템은 점점 더 에이전트 기반으로 진화하고 있다. 단순히 대시보드를 감시하는 수준을 넘어, 스스로 사고하고 계획하며 인프라 내 변경을 실행하는 자율형 에이전트가 배치되고 있다. 이들 에이전트는 PDF 문서를 읽을 수 없고, 법률 용어로 작성된 보안 정책의 취지를 해석하지도 못한다. 자율형 IT 시대에 통제력을 유지하려면 고정된 규칙에 머무르지 않고 ‘에이전트 헌법’, 즉 앤트로픽이 제시한 ‘헌법 중심 AI(Constitutional AI)’를 기업 환경에 적용해야 한다. 이는 AI의 문제점을 AI가 스스로 검증하고 고치기 위한 시스템을 의미한다.

문서 속 정책에서 코드로 구현된 정책으로

과거 IT 거버넌스는 사후 대응만 가능한 ‘체크리스트’ 방식이었다. 그러나 오늘날 기업은 정책을 코드로 구현하는 ‘PaC(Policy as Code)’로의 전환이 필요하다.

• 전전두엽 역할: 에이전트 헌법은 자율 시스템 기계가 읽을 수 있는 기본 원칙 집합이다.
• 운영 경계: 에이전트가 수행할 수 있는 작업 범위와 절대 넘지 말아야 할 윤리적 한계를 규정한다.
• 실행 가능한 규칙: 코드로 구현된 강제 규칙의 예로는 ‘피크 시간대에는 인간 개입 토큰 없이는 운영 데이터베이스를 수정하지 않는다’와 같은 원칙이 있다.
• LLM 이해 가능성: 이러한 규칙은 실행 가능하며, 오케스트레이션을 담당하는 LLM이 이해하고 활용할 수 있다.

이런 전환은 근본적인 변화를 의미한다. IT 전문가의 역할은 ‘운영자’에서 ‘의도 설계자’로 변화하고 있다. IT 직원은 더 이상 시스템을 직접 조작하는 사람이 아니라, 자율 시스템이 따라야 할 행동 규칙을 설계하는 주체가 되고 있다.

IT 운영을 위한 자율성 계층 구조

기업이 ‘킬 스위치’에 대한 통제권을 유지하면서 AI 역량을 확장하려면, ‘자율성의 계층 구조’에 초점을 맞출 필요가 있다. 이는 1978년 연구자 토머스 셰리던와 윌리엄 버플랭크의 기초 연구에서 제시된 프레임워크에 기반한 개념이다.

1단계: 완전 자율화 영역(가장 쉽게 도입할 수 있는 영역)

• 이는 사람이 개입하는 비용이 해당 작업의 가치보다 더 큰 업무를 의미한다.
• 사례
  • 자동 확장
  • 로그 로테이션
  • 기본 티켓 라우팅
  • 캐시 정리
• 거버넌스: 사전에 정의된 임계값 조건에 따라 동작하는 통제된 자동화 영역(sandbox of trust)에서 관리된다.

2단계: 감독형 자율화 영역(사전 확인 구간)

• 에이전트가 데이터 수집과 문제 원인 분석, 해결 방안 도출 등 핵심 작업을 수행하지만, 최종 실행 단계에서는 사람의 승인, 즉 확인 절차가 필요한 단계다.
• 사례
  • 시스템 패치
  • 사용자 계정 프로비저닝
  • 비중요 설정 변경
• 거버넌스: 에이전트는 해당 조치를 왜 수행하려는지에 대한 판단 근거, 즉 추론 과정을 관리자에게 제시해야 한다.

3단계: 사람 전용 영역

• 어떤 상황에서도 에이전트가 자율적으로 수행해서는 안 되는, 시스템의 존립과 직결된 핵심 작업을 의미한다.
• 사례
  • 데이터베이스 삭제
  • 핵심 보안 설정 우회
  • 에이전트 헌법 자체에 대한 수정
• 거버넌스: 다단계 인증(MFA) 또는 복수 인원의 이중 승인과 같은 강력한 통제 절차를 적용한다.

숨겨진 공격 표면 줄이기

중앙화된 헌법 체계를 구현하면, 중앙 IT의 관리 및 감독 없이 배포되는 섀도우 AI 에이전트로 인한 리스크를 완화할 수 있다.

• 통합 API: 모든 에이전트는 핵심 인프라와 상호작용하기 전에 해당 운영 원칙 체계에 따라 인증을 거쳐야 한다.
• 컴플라이언스 이력: 이를 통해 SOC2나 EU AI 법과 같은 규제 대응에 활용할 수 있는 중앙화된 감사 이력이 생성된다.
• 검증 가능한 의사결정: 자율적으로 내려진 판단과 실행에 대한 검증 가능한 기록을 축적할 수 있다.

기계 중심 세계 속 사람의 목소리

이른바 ‘헌법’은 코드가 아니라, 엔지니어의 경험과 판단이 집약된 사람의 문서다. 따라서 사람의 역할은 여전히 중요하다.

• 의도 설계자: IT 전문가의 역할은 ‘운영자’에서 ‘의도의 설계자’로 변화하고 있다.
• 문화적 전환: IT 팀은 개인이 나서서 문제를 해결하는 방식에서 벗어나, 시스템 중심의 거버넌스 문화로 전환해야 한다.

‘헌법 제정 회의’를 시작할 때

2020년대 후반에도 PDF 형식의 기존 SOP에 의존한다면, IT 운영은 비즈니스의 발목을 잡는 병목으로 전락할 가능성이 크다.

지금 바로 취해야 할 단계는 다음과 같다.

• 레드라인 정의: 수석 아키텍트를 중심으로 3단계 영역의 경계를 명확히 설정한다.
• 자동화 성과 도출: 즉시 자동화가 가능한 1단계 업무를 식별한다.
• 전략에 집중: 사람은 봇을 감시하는 데 시간을 쓰는 대신, 전략과 혁신에 집중하도록 방향을 전환한다.

dl-ciokorea@foundryco.com

For decades, the IT operations manual was a dense, 50-page PDF — a document designed by humans, for humans, and usually destined to gather digital dust until an audit required its retrieval. But as we enter 2026, the traditional standard operating procedure (SOP) is officially on life support. Humans are no longer the primary users of their own manuals.

Our systems are becoming agentic, deploying autonomous agents that don’t just monitor dashboards but actively “think,” plan, and execute changes within our infrastructure. These agents cannot read a PDF, nor can they “interpret the spirit” of a security policy written in legalese. If you want to maintain control in an era of autonomous IT, you must move beyond static guardrails and adopt an Agentic Constitution, which is the enterprise application of Constitutional AI, a term pioneered by Anthropic.

From policy on paper to policy as code 

In the past, IT governance was a reactive “check-the-box” exercise. The modern enterprise must shift toward Policy as Code (PaC).

• The pre-frontal cortex: An Agentic Constitution is a machine-readable set of foundational principles for your autonomous systems.
• Operational boundaries: They define what an agent can do and the ethical boundaries it must never cross.
• Actionable rules: An example of an encoded hard rule is: “Never modify production data during peak hours without a human-in-the-loop token”.
• Understandable by LLMs: These rules are actionable and understandable by the models powering your orchestration.

This shift represents a fundamental transformation: the role of the IT professional is moving from “Operator” to “Architect of Intent”. IT professionals are no longer the ones turning the wrenches; they are the ones writing the rules of engagement.

The hierarchy of autonomy: A framework for IT ops 

To scale AI capabilities without ceding total control of the “kill switch”, enterprises should adopt a hierarchy of autonomy, a framework credited to the foundational work of Thomas Sheridan & William Verplank (1978).

Tier 1: Full autonomy (the low-hanging fruit) 

• Description: Tasks where the cost of human intervention exceeds the value of the task.
• Examples: 
  • Auto-scaling 
  • Log rotation 
  • Basic ticket routing 
  • Cache clearing 
• Governance: Defined by threshold-based triggers within a “sandbox of trust”.

Tier 2: Supervised autonomy (the ‘check-back’ zone) 

• Description: Agents perform heavy lifting — gathering data and identifying fixes — but require a “human nod” before final execution.
• Examples: 
  • System patching 
  • User provisioning 
  • Non-critical configuration changes 
• Governance: Agents must present a “reasoning trace” to the admin explaining why the action is being taken.

Tier 3: Human-only (the red line) 

• Description: “Existential” actions that no agent should ever perform autonomously.
• Examples: 
  • Database deletions 
  • Critical security overrides 
  • Modifications to the Agentic Constitution itself 
• Governance: Multi-factor authentication (MFA) or multi-person “dual-key” approvals.

Reducing the ‘hidden attack surface’ 

Implementing a centralized constitution helps mitigate the risks of shadow AI agents — autonomous tools deployed without central IT oversight.

• Unified API: Any agent must “authenticate” against the constitution before it can interact with core infrastructure.
• Compliance history: This creates a centralized audit trail invaluable for compliance frameworks like SOC2 or the EU AI Act.
• Verifiable decision-making: You are building a verifiable history of autonomous decision-making.

The human voice in a machine world 

The “Constitution” is a human document representing the collective wisdom of your engineers.

• Architects of intent: The role of the IT professional shifts from “Operator” to “Architect of Intent”.
• Cultural shift: IT teams must move away from “hero culture” firefighting toward a culture of systemic governance.

Conclusion: Starting your constitutional convention 

If you rely on human-readable SOPs in the second half of the decade, your IT operations will become a bottleneck for the business.

Steps to take this quarter:

• Identify red lines: Gather lead architects to define your Tier 3 boundaries.
• Map automated wins: Identify Tier 1 tasks for immediate automation.
• Focus on strategy: Ensure humans focus on strategy and innovation, not babysitting a bot.

This article is published as part of the Foundry Expert Contributor Network.
Want to join?

AI won’t replace people. But leaders who ignore workforce redesign will begin to fail and be replaced by leaders who adapt and quickly.

For the last decade or so, digital transformation has been framed as a technology challenge. New platforms. Cloud migrations. Data lakes. APIs. Automation. Security layered on top. It was complex, often messy and rarely finished — but the underlying assumption stayed the same: Humans remained at the center of work, with technology enabling them.

AI breaks that assumption.

Not because it is magical or sentient — it isn’t — but because it behaves in ways that feel human. It writes, reasons, summarizes, analyzes and decides at speeds that humans simply cannot match. That creates a very different emotional and organizational response to any technology that has come before it.

I was recently at a breakfast session with HR leaders where the topic was simple enough on paper: AI and how to implement it in organizations. In reality, the conversation quickly moved away from tools and vendors and landed squarely on people — fear, confusion, opportunity, resistance and fatigue. That is where the real challenge sits.

AI feels human and that changes everything

AI is just technology. But it feels human because it has been designed to interact with us in human ways. Large language models combined with domain data create the illusion that AI can do anything. Maybe one day it will. Right now, what it can do is expose how unprepared most organizations are for the scale and pace of change it brings.

We are all chasing competitive advantages — revenue growth, margin improvement, improving resilience — and AI is being positioned as the shortcut. But unlike previous waves of automation, this one does not sit neatly inside a single function.

Earlier this year I made what I thought was an obvious statement on a panel: “AI is not your colleague. AI is not your friend. It is just technology.” After the session, someone told me — completely seriously — that AI was their colleague. It was listed on their Teams org chart. It was an agent with tasks allocated to it.

That blurring of boundaries should make leaders pause.

Perception becomes reality very quickly inside organizations. If people believe AI is a colleague, what does that mean for accountability, trust and decision-making? Who owns outcomes when work is split between humans and machines? These are not abstract questions — they show up in performance, morale and risk.

When I spoke to younger employees outside that HR audience, the picture was even more stark. They understood what AI was. They were already using it. But many believed it would reduce the number of jobs available to their generation. Nearly half saw AI as a net negative force. None saw it as purely positive.

That sentiment matters. Because engagement is not driven by strategy decks — it is driven by how people feel about their future.

Roles, skills and org design are already out of date

One of the biggest problems organizations face is that work is changing faster than their structures can keep up.

As Zoe Johnson, HR director at 1st Central, put it: “The biggest mismatch is in how fast the technology is evolving and how possible it is to redesign systems, processes and people impacts to keep pace with how fast work is changing. We are seeing fast progress in our customer-facing areas, where efficiencies can clearly be made.”

Job frameworks, skills models and career paths are struggling to keep up with reality. This mirrors what we are now seeing publicly, with BBC reporting that many large organizations expect HR and IT responsibilities to converge as AI reshapes how work actually flows through the enterprise.

AI does not neatly replace a role — it reshapes tasks across multiple roles simultaneously. That shift is already forcing leadership teams to rethink whether work should be organized by function at all or instead designed end‑to‑end around outcomes. That makes traditional workforce planning dangerously slow.

Organizations are also hitting change saturation. We have spent years telling ourselves that “the only constant is change,” but AI feels relentless. It lands on top of digital transformation, cloud, cyber, regulation and cost pressure.

Johnson is clear-eyed about this tension: “This is a constant battle, to keep on top of technology development but also ensure performance is consistent and doesn’t dip. I’m not sure anyone has all the answers, but focusing change resource on where the biggest impact can be made has been a key focus area for us.”

That focus is critical. Because indiscriminate AI adoption does not create advantages — it creates noise.

This is no longer an IT problem

For years, organizations have layered technology on top of broken processes. Sometimes that was a conscious trade-off to move faster. Sometimes it was avoidance. Either way, humans could usually compensate.

AI does not compensate. It amplifies. This is the same dynamic highlighted recently in the Wall Street Journal, where CIOs describe AI agents accelerating both productivity and structural weakness when layered onto poorly designed processes.

Put AI on top of a poor process and you get faster failure. Put it on top of bad data and you scale mistakes at speed. This is not something a CIO can “fix” alone — and it never really was.

The value chain — how people, process, systems and data interact to create outcomes — is the invisible thread most organizations barely understand. AI pulls on that thread hard.

That is why the relationship between CIOs and people leaders has moved from important to existential.

Johnson describes what effective partnership actually looks like in practice: “Constant communication and connection is key. We have an AI governance forum and an AI working group where we regularly discuss how AI interventions are being developed in the business.”

That shared ownership matters. Not governance theatre, but real, ongoing collaboration where trade-offs are explicit and consequences understood.

Culture plays a decisive role here. As Johnson notes, “Culture and trust is at the heart of keeping colleagues engaged during technological change. Open and honest communication is key and finding more interesting and value-adding work for colleagues.”

AI changes what work is. People leaders are the ones who understand how that lands emotionally.

The CEO view: Speed, restraint and cultural expectations

From the CEO seat, AI is both opportunity and risk. Hayley Roberts, CEO of Distology, is pragmatic about how leadership teams get this wrong.

“All new tech developments should be seen as an opportunity,” she said. “Leadership is misaligned when the needs of each department are not congruent with the business’s overall strategy. With AI it has to be bought in by the whole organization, with clear understanding of the benefits and ethical use.”

Some teams want to move fast. Others hesitate — because of regulation, fear or lack of confidence. Knowing when to accelerate and when to hold back is a leadership skill.

“We love new tech at Distology,” Roberts explains, “but that doesn’t mean it is all going to have a business benefit. We use AI in different teams but it is not yet a business strategy. It will become part of our roadmap, but we are using what makes sense — not what we think we should be using.”

That restraint is often missing. AI is not a race to deploy tools — it is a race to build sustainable advantage.

Roberts is also clear that organizations must reset cultural expectations: “Businesses are still very much people, not machines. Comprehensive internal assessment helps allay fear of job losses and assists in retaining positive culture.”

There is no finished AI product. Just constant evolution. And that places a new burden on leadership coherence.

“I trust what we are doing with our AI awareness and strategy,” Roberts says. “There is no silver bullet. Making rash decisions would be catastrophic. I am excited about what AI might do for us as a growing business over time.”

Accountability doesn’t disappear — it concentrates

One uncomfortable truth sits underneath all of this: AI does not remove accountability. It concentrates it. Recent coverage in The HR Director on AI‑driven restructuring, role redesign and burnout reinforces that outcomes are shaped less by the technology itself and more by the leadership choices made around design, data and pace of change.

When decisions are automated or augmented, the responsibility still sits with humans — across the entire C-suite. You cannot outsource judgement to an algorithm and then blame IT when it goes wrong.

This is why workforce redesign is not optional. Skills, org design and leadership behaviors must evolve together. CIOs bring the technical understanding. CPOs and HRDs bring insight into capability, culture and trust. CEOs set the tone and pace.

Ignore that partnership and AI will magnify every weakness you already have.

Get it right and it becomes a powerful force for growth, resilience and better work.

The workforce shift is already underway. The question is whether leaders are redesigning for it — or reacting too late.

This article is published as part of the Foundry Expert Contributor Network.
Want to join?

Modern CIOs and tech leaders carry responsibility not only for an organization’s technology but, as key partners, for its entire business success. So having access to readily transferable lessons is critical in order to solve real business challenges, and lead with clarity, confidence, and purpose.

As a jumping off point, I’ve distilled here some of my favourite maxims from different business functions.

Maxim 2: Try to be human

You’re more interesting than you think. Try to be human. I realize this is a tough ask for us classic IT introvert types, but with many interactions now conducted remotely, it’s even more important to find opportunities to meet in person.

Letting people know what makes you tick personally is of more interest than you could probably imagine. Colleagues are interested in you as a whole person, not simply as the person they work with. So don’t be afraid to bring yourself to work, as the phrase goes. This allows others to do the same, and to talk about their own feelings and circumstances.

As an INTP (an introverted, intuitive, thinking, and perceiving type from the Myers-Briggs personality assessment), social events aren’t my natural environment. And we’ve probably all experienced how work and socializing sometimes don’t mix. Is an orchestrated corporate event all that comfortable for anyone? But try to show up and meet people, relax a bit, and have some fun.

Maxim 6: Beware the IT cultural cringe

IT people often prefer to vent about the technology-ignorant business rather than stand up and explain the tech. Instead of declaring something’s bad for the company or a dead-end, they shrug and say the business just doesn’t get it.

No matter how great your strategy is, your plans will fail without a company culture that encourages people to implement it. I know from speaking to other CIOs that a frequent role for them is standing up for IT and defending their teams in a culture where the business blames IT for its failures.

It’s therefore vital to coach your teams to deal on equal terms with their internal business customers. Key to this is talking in business terms, not IT jargon. The reason for not adopting a nonstandard piece of tech is it’ll inflate future company running costs, not that it doesn’t neatly fit the IT estate. So stand up and be counted on a matter of tech principle, and win the debate.

Maxim 8: There are no IT projects, only business projects.

When IT projects fail, it’s often because of a lack of ownership by the business.

The entire purpose of your IT department is to move the organization forward. So any investment must deliver on quantifiable financial targets or defined business objectives. If it doesn’t, move on. This is fundamental. Forgetting to do so is easy when under pressure, as others press you with their own agendas, but dangerous for you and the business.

Everything I’ve learned and seen reinforces this. Without this focus, you’re just an IT supplier taking orders, not the executive IT partner of the business. Question any actions by your team that can’t be linked back to the company’s core objectives.

It all comes down to building relationships based on trust with your business colleagues who recognize that you understand what the business needs and can afford, so challenge projects not owned by the business leaders.

Maxim 10: The CIO as the personification of IT

Be vocal about your team’s successes and be honest about your mistakes. As CIO, you’re the face of the IT function in your organization, and you set the tone for everyone in IT.

Try not to talk about the business and IT as separate entities. You and your team are just as integral to the company as sales, operations, or finance. Always talk about our business needs and what we should do.

Remember, you’re accountable for all the IT. These days, we talk about being authentic, so being honest about your slip-ups, and how you feel about them, is important in establishing your reputation, both internally and externally.

Explain a success to others in the organization and why it worked. Bring out how collaboration between their teams and IT, working to aligned plans and objectives, made good things happen for everyone involved.

Maxim 36: Join up digital and IT

Digital natives need to work together with old techies. Advances of the last decade have been delivered by fast-moving digital startups, financed by deep-pocketed investors. Unsurprisingly, this has spawned organizational impatience with the costs and time taken by traditional or legacy IT functions. This frustration can then translate into setting up a completely separate digital department under a CDO, charged with implementing the new and faster-moving business.

Your current business is built on long-established ways of working, and processes that remain necessary, unless you’re going to build them all a second time for the new digital channel. If not, then new components, including services and products, will have to interface with existing systems, as well as firmly established and mission-critical business processes. So with this dynamic, ensure that both traditional IT and new digital report to you.

Maxim 56: AI is a tech-driven business revolution

AI is the most overhyped bandwagon in technology, more than bitcoin, big data, and augmented and virtual reality. Nevertheless, it’s the most far-reaching tech-driven change since the advent of the internet. In a matter of months, AI and AI agents are doing to white-collar jobs what production line robots did to blue-collar jobs 20 years ago.

AI is transforming the world and we’re just at the beginning of this revolution. So what are you doing about it?

Your challenge as CIO is that AI has cut through to your board and executive leadership like nothing before. Furthermore, all your partners and suppliers are building AI agents into their software and services. Plus, all your best digital innovators in the business, and definitely all your recent grad hires, are using Chat GPT and bespoke AI tools in their day jobs. As CIO, you hold the keys to AI working well by effectively wielding the data in your systems. After all, you and your team are the ones who best understand how the AI works as the means to achieve business value.

As 2026 begins, my recent conversations with Chief Information Officers across Southeast Asia provided me with a grounded view of how digital transformation is evolving. While their perspectives differ in nuance, they converge on several defining shifts: the maturation of artificial intelligence, the emergence of autonomous systems, a renewed focus on data governance, and a reconfiguration of work. These changes signal not only technological advancement but a rethinking of how Southeast Asia organizations intend to compete and create value in an increasingly automated economy.

For our CIOs, the year ahead represents a decisive moment as AI moves beyond pilots and hype cycles. Organizations are expected to judge AI by measurable business outcomes rather than conceptual promise. AI capabilities will become standard features embedded across applications and infrastructure, fundamental rather than differentiating. The real challenge is no longer acquiring AI technology but operationalizing it in ways that align with strategic priorities.

Among the most transformative developments is the rise of agentic AI – autonomous agents capable of performing tasks and interacting across systems. CIOs anticipate that organizations will soon manage not a single AI system but networks of agents, each with distinct logic and behaviour. This shift ushers in a new strategic focus, agentic AI orchestration. Organizations will need platforms that coordinate multiple agents, enforce governance, manage digital identity, and ensure trust across heterogeneous technology environments. As AI ecosystems grow more complex, the CIO’s role evolves from integrator to orchestrator who directs a diverse array of intelligent systems.

As AI becomes more central to operations, data governance emerges as a critical enabler. Technology leaders expect 2026 to expose the limits of weak data foundations. Data quality, lineage, access controls, and regulatory compliance determine whether AI initiatives deliver value. Organizations that have accumulated “data debt” will be unable to scale, while those that invest early will move with greater speed and confidence.

Automation in physical environments is also set to accelerate as CIOs expect robotics to expand across healthcare, emergency services, retail, and food and beverage sectors. Robotics will shift from specialised deployments to routine service delivery, supporting productivity goals, standardizing quality, and addressing persistent labour constraints.

Looking ahead, our region’s CIOs point to the early signals of quantum computing’s relevance. While still emerging, quantum technologies are expected to gain visibility through evolving products and research. In my view, for Southeast Asia organizations, the priority is not immediate adoption but proactive monitoring, particularly in cybersecurity and long-term data protection, without undertaking premature architectural shifts.

Perhaps the most provocative prediction concerns the nature of work. As specialised AI agents take on increasingly complex task chains, one CIO anticipates the rise of “cognitive supply chains” in which work is executed largely autonomously. Traditional job roles may fragment into task-based models, pushing individuals to redefine their contributions. Workplace identity could shift from static roles to dynamic capabilities, a broader evolution in how people create value in an AI-native economy.

One CIOs spotlight the changing nature of software development where natural-language-driven “vibe coding” is expected to mature, enabling non-technical teams to extend digital capabilities more intuitively. This trend will not diminish the relevance of enterprise software as both approaches will coexist to support different organizational needs.

CIO ASEAN Editorial final take:

Collectively, these perspectives shared by Southeast Asia’s CIO community point to Southeast Asia preparing for a structurally different digital future, defined by embedded AI, scaled autonomous systems, and disciplined data practices. The opportunity is substantial, but so is the responsibility placed on technology leaders.

As 2026 continue to unfold, the defining question will not simply be who uses AI, but who governs it effectively, integrates it responsibly, and shapes its trajectory to strengthen long-term enterprise resilience. Enjoy reading these top predictions for 2026 by our region’s most influential CIOs who are also our CIO100 ASEAN & Hong Kong Award 2025 winners:

Ee Kiam Keong Deputy Chief Executive (Policy & Development) concurrent Chief Information Officer InfoComm Technology Division Gambling Regulatory Authority Singapore   Prediction 1 AI continue to lead its edge esp. Agentic AI would be getting more popular and used, and AI Governance in terms AI risks and ethnics would get more focused   Prediction 2 Quantum Computing related products should start to evolve and more apparent.   Prediction 3 Deployment of robotic applications would be widened esp. in medical, emergency response and casual activities such retail, and food and beverage etc.

Ng Yee Pern, Chief Technology Officer Far East Organization   Prediction 4 AI deployments will start to mature, as enterprises confront the disconnect between the inflated promises of AI vendors and the actual value delivered.   Prediction 5 Vibe coding will mature and grow in adoption, but enterprise software is not going away. There is plenty of room for both to co-exist.

Athikom Kanchanavibhu Executive Vice President, Digital & Technology Transformation & Chief Information Security Officer Mitr Phol Group   Prediction 6 The Next Vendor Battleground: Agentic AI Orchestration By 2026, AI will no longer be a differentiator, it will be a default feature, embedded as standard equipment across modern digital products. As every vendor develops its own Agentic AI, enterprises will manage not one AI, but an orchestra of autonomous agents, each optimized for its own ecosystem.   The new battleground will be Agentic AI Orchestration where platforms can coordinate, govern, and securely connect agentic AIs across vendors and domains. 2026 won’t be about smarter agents, but about who can conduct the symphony best-safely, at scale, and across boundaries.   Prediction 7 Enterprise AI Grows Up: Data Governance Takes Center Stage 2026 will mark the transition from AI pilots to AI in production. While out-of-the-box AI will become common, true competitive advantage will come from applying AI to enterprise-specific data and context. Many organizations will face a sobering realization: AI is only as good as the data it is trusted with.   As AI moves into core business processes, data governance, management, and security will become non-negotiable foundations. Data quality, access control, privacy, and compliance will determine whether AI scales or stalls. In essence, 2026 will be the year enterprises learn that governing data well is the quiet superpower behind successful AI.

Jackson Ng Chief Technology Officer and Head of Fintech Azimut Group   Prediction 8 In 2026, organizations will see AI seeking power while humans search for purpose. Cognitive supply chains of specialized AI agents will execute work autonomously, forcing individuals to redefine identity at work, in service, and in society. Roles will disintegrate, giving way to a task-based, AI-native economy

IT 업계의 분위기만 보면, AI 에이전트가 전체 비즈니스 프로세스를 자동화해 전 세계 기업을 바꿔놓을 것 같다. 하지만 현실은 전혀 다르다.

와튼 스쿨과 지비케이 컬렉티브(GBK Collective)가 발표한 AI 도입 조사에 따르면, 기업 IT 의사결정권자 58%가 조직 내에서 AI 에이전트를 시범 운영하고 있다고 답했으며, 대다수는 프로세스 자동화, 워크플로우 효율화, 고객 서비스 등의 사용례에 적용할 계획이라고 밝혔다.

하지만 어디까지나 파일럿 단계이며, 프로덕션 환경에 적용된 것은 아니다. 완성된 인간-AI 에이전트 협업 워크플로우를 위한 교본 같은 것은 아직 없다.

IT 부서가 AI를 사용해 운영을 자동화할 최선의 길을 두고 고민하는 과정에서 HR과의 긴밀한 협력이 필수적인 요소로 떠올랐다. IT와 HR의 협력은 혼란을 최소화하고 인간과 AI가 동료로서 같이 근무하는 환경이 현실화될 때 등장할 새 역할, 프로세스, 팀 구조에 기업이 빠르게 적응할 수 있도록 지원한다.

AI 에이전트를 조직 안으로 들이는 방법

사이버 보안 회사 소포스(Sophos)의 CIO 토니 영은 책임 있는 AI 배포를 위한 변화관리에서 IT와 HR의 밀착이 중요하다고 강조했다. 영은 “올바른 접근은 HR 전문가들과 함께하고, 어떻게 구성원을 변화에 동참시킬지 이해하는 것”이라며 HR과의 동행을 핵심으로 짚었다.

영이 그리는 미래 조직에는 자동화 전문가뿐 아니라, 콘텐츠를 큐레이션하고 데이터를 다뤄 에이전틱 AI 전환을 매끄럽게 만드는 인력도 늘어난다. HR은 이런 ‘새로 생겨나는 전문 인력’을 팀에 자연스럽게 섞어 배치하는 역할을 맡는다.

AI 전환기에는 직원의 불안을 완화할 ‘작은 장치’도 유용하다. 소포스의 마케팅 부서는 AI 에이전트를 조직도에 팀 구성원으로 포함했고, 새 에이전트가 투입되면 사람과 마찬가지로 ‘신규 팀원 소개’ 공지를 낸다. IT 서비스 데스크에는 사람 직원이 디지털 동료와 성과를 비교할 수 있는 리더보드도 마련했다. 다만 사람 직원이 AI 에이전트의 결과를 검증하고 감시하는 구조를 유지해 휴먼 인 더 루프(human-in-the-loop) 베스트 프랙티스에 맞춘 운영을 이어간다는 설명이다.

역량 기준도 달라진다. 영은 “LLM을 활용하는 법이나 에이전트를 만드는 법은 엑셀을 다루는 것과 비슷하다”라며, “모두가 갖춰야 하는 새로운 기본 역량이 됐다”라고 표현했다. 이를 위해 CIO가 HR 리더와 함께 AI 교육 의제를 정하고, 생성형 AI 관련 자격·교육과정까지 포함한 ‘전사 학습 로드맵’을 마련해야 한다는 주장이다.

수백 수천의 봇이 일하는 회사의 운영모델

에이전틱 AI 환경을 완성한 기업은 어떤 모습일까. 수백 수천의 자율 ‘봇’이 엔드 투 엔드 업무 프로세스를 수행하고, 이들 봇을 관리하며 작업 이탈을 막는 ‘보스 봇’이 상단에 있는 구조가 거론된다. 인간 조직이 지식 노동을 해온 방식과 대칭적이지만, 실제로는 에이전트와 함께 일하기 위한 새로운 운영모델이 필요하다는 것이 핵심이다.

앞으로 IT 부서는 에이전트의 의사결정 체계와 그에 따른 워크플로우를 설계·배포·운영하는 책임이 커질 것이다. 워크플로우는 부서마다 다르게 설계돼야 한다. 예를 들어 콜센터 업무를 AI로 자동화하려면, 사람이 에이전트를 모니터링하는 법을 배워야 한다. 이는 기존 상담 인력의 업무 도구를 넘어서는 관리·기술 역량을 요구한다.

기업 맥킨지의 수석 파트너 클레멘스 야르타르는 “통화 의도를 이해하고 경계를 설정하는 등 새로운 역량이 필요하다”라고 설명했다. 기존의 ‘사람 중심’ 운영 방식에 익숙한 조직일수록 프로세스 관리 역량을 새로 길러야 한다는 지적이다.

세일즈·마케팅에 AI 에이전트를 접목할 때는 CRM 등 고객 접점 시스템 전반의 워크플로우 재설계가 과제로 떠오른다. 운영 조직 등 다른 기능도 마찬가지다. 어떤 업무 흐름이든 HR은 명확하고 일관된 커뮤니케이션으로 직원들이 받을 충격을 줄이고, IT와 각 부서가 재교육·리스킬링 전략을 실행하도록 돕는 역할을 맡을 수 있다.

마이크로소프트는 IT와 HR이 인간·디지털 인력을 함께 조율하는 CRO(Chief Resource Officer)’ 같은 새 역할을 만들 수 있다고 전망했고, 일부 조직에는 ‘에이전트 보스(agent boss)’가 등장할 가능성도 제기했다. 맥킨지는 AI 윤리·책임 사용, AI 품질 보증 책임, 에이전트 코치 등 신규 직무가 생겨날 것으로 내다봤다.

데이터 신뢰성은 자율성의 생명줄

과제가 어렵지만, 넘지 못할 수준은 아니라는 평가도 나온다. 다만 에이전틱 AI 아키텍처에 지나친 의사결정 권한을 부여하는 것은 위험하다는 지적이 적지 않다.

핀옵스 플랫폼 서비스 업체 두잇(DoiT)의 필드 CTO 아미트 킨하는 서로 다른 플랫폼 전반에 걸친 기술적 난제와 ‘암묵지(tribal knowledge)’의 공백을 리스크로 지목했다. 예를 들어, 저연차 개발자에게 업무를 주면 필요할 때 선임에게 도움을 구할 수 있지만, 현재 AI 에이전트는 그런 지식 접근 메커니즘이 부족하다는 지적이다. 킨하는 “진실의 출처(source of truth)가 어디인지가 중요하다”며 “그 기반이 유효하지 않으면 의사결정 트리 전체가 무효가 된다”라고 우려를 드러냈다.

에이전트 실행의 파급력도 크다. 예컨대 15개 시스템에 걸쳐 업데이트할 수 있는 멀티 에이전트 시스템은 하위 단계에 연쇄적인 영향을 주며 수익성에 실질적 타격을 입힐 수 있다. 이런 리스크를 줄이려면 거버넌스 차원에서 ‘체크포인트’를 두는 방식이 대안으로 꼽힌다. 어떤 에이전트는 단독 결정을 허용하되, 다른 에이전트는 사람 승인 없이는 실행하지 못하게 하는 식이다.

킨하는 “완성하기 가장 어려운 부분이 의사결정 자율성”이라며, 자율성이 너무 낮으면 사람에게 매번 확인하느라 자동화가 멈추고, 너무 높으면 치명적 실수가 발생할 수 있다고 설명했다. 또, 목표와 의도를 명확히 하고, 데이터 위생을 확고히 해야 한다고 덧붙였다.

밝지만 예측할 수 없는 미래

기술과 프로세스의 퍼즐이 맞춰지기 시작하면, 사람에서 ‘사람+기계’로 일하는 방식으로 넘어가는 전환을 돕는 축은 결국 IT와 HR의 파트너십이 될 가능성이 크다. AI 에이전트를 도입하는 모든 기업은 프로세스 실행 방식과 성과 측정 체계를 더 의도적으로 설계해야 한다.

야르타르는 “각 기능 영역의 구성원 모두가 의도 설정, 경계 설정, 측정 역량을 끌어올려야 한다”라며 “그 과정에는 여러 해가 걸릴 것”이라고 내다봤다. 영도 “기업마다 속도가 다를 것”이라며, 자동화를 강하게 밀어붙이는 조직과 그렇지 않은 조직 사이에 새로운 ‘격차’가 생길 수 있다고 덧붙였다.

한 가지, 직장에서 인간과 기계가 함께 일하는 환경과 관련된 도전은 이제 막 시작됐다는 점은 분명하다.
dl-ciokorea@foundryco.com

Tech industry sentiment suggests that AI agents will automate entire business processes, potentially transforming companies worldwide.

Today’s reality is starkly different.

Fifty-eight percent of enterprise IT decision-makers say their organizations are piloting AI agents, with the majority targeting process automation, workflow efficiencies, or customer service, among other use cases, according to AI adoption research published by Wharton and the GBK Collective.

Again, these are pilots — not production implementations. There isn’t yet a playbook for fully baked human-AI agent workflows.

Still, as IT departments wrestle with the best path forward for using AI to automate operations, close partnership with human resources departments will be essential to minimize disruption and ensure the organization is primed to capitalize on the new roles, processes, and team structures that will arise as true human-AI coworking arrives.

Bringing AI agents into the fold

Tight interaction between IT and HR is crucial for the change management required for responsible AI deployment, says Sophos CIO Tony Young, who is spearheading the deployment of AI at the MDR vendor, including Microsoft Copilot. “The right approach is engaging with your HR pros and understanding how we bring the workforce along,” Young says.

For example, Young envisions more companies will employ automation experts, along with those who understand how to curate content and work with data to smooth the transition to agentic AI. HR can help blend the budding array of specialists.

Moreover, a little anthropomorphization can go a long way toward easing the transition to digital colleagues, Young adds.

The marketing organization at Sophos now includes AI agents in org charts as part of its teams, working alongside humans. New agents get new team member announcements — just like humans, says Young.

And Sophos’ IT service desk function now features a leaderboard that allows humans to see how they stack up against their digital coworkers. Human staffers monitor the AI agents to validate their work, consistent with human-in-the-loop best practices.

“Understanding how to use an LLM, or how to create an agent is like mastering Excel,” Young says. “That’s a new baseline skill that we all need to have.”

To get there, CIOs need to partner with HR leaders to help set the workforce AI training agenda, which could include emerging gen AI certifications as well as coursework for driving AI change.

What the agent-infused organization of the future will look like

What will fully agentic businesses look like in the future? Picture hundreds or thousands of autonomous “bots” working together to facilitate the execution of business processes end-to-end. These worker bots will likely be managed by a “boss” bot that ensures they stay on task.

If this sounds familiar it’s because it’s a symmetrical analogy for how humans have long performed knowledge work.

Yet organizations require a new operating model for working with agents. It will be incumbent on IT departments to stage and manage agent decision trees and the resulting workflows. These workflows will vary by function.

For instance, organizations that choose to automate call center operations with AI will need to train humans to monitor agents — a managerial and technical skill that goes beyond most call center associates’ current toolboxes.

“It requires a new skillset, including understanding the intent of calls and setting boundaries,” says Klemens Hjartar, senior partner at McKinsey. This requires new process management muscles for organizations accustomed to working a certain, human-centric way.

The introduction of AI agents to sales and marketing processes presents different challenges involving various workflows for CRM and other systems of engagement. The same can be said for operations teams and other functions likely to be impacted by agentic AI.

Whatever the workflow, HR can help soften the impact on teams through clear, consistent communication, as well as messaging around how IT and other departments can reskill their teams for the new era.

Microsoft predicted that IT and HR teams will forge new roles such as chief resource officers to help balance human and digital workers, while some organizations may install “agent bosses.” McKinsey envisions new roles for AI ethics and responsible usage, AI quality assurance leads, and agent coaches.

The hurdles are huge but not insurmountable

In short, wholesale changes to organizational dynamics are on the horizon, with IT and HR serving on the front lines of these transformations — mostly in tandem.

While these changes are a ways away, most organizations aren’t ready for it — but need to keep this future in mind as they plan their way forward.

One challenge is the fact that allocating too much decision-making authority to agentic AI architectures poses significant risks, due to technical challenges across disparate platforms and implicit knowledge gaps, says Amit Kinha, field CTO of FinOps platform provider DoiT.

For example, if you give a junior programmer some tasks to accomplish, they can turn to more experienced engineers when they need help. Today there isn’t a mechanism for AI agents to access the same tribal knowledge, Kinha says.

“Where is the source of truth coming from?” Kinha wonders. “Because if it’s not valid the whole decision tree will be invalid as well.” 

The ramifications of agentic actions loom large. A multi-agent system with the power to update across 15 systems could have significant impacts downstream that materially impact the bottom line, Kinha says.

One approach may include instituting checkpoints as part of organizational governance strategies. For instance, while some AI agents may be authorized to make individual decisions, others may have to seek approval from a human.

“The hardest part to master is decision autonomy,” Kinha says. Agents with too little autonomy will regularly check with humans, stunting automation. Those with too much will make mistakes that could be catastrophic. In addition to being explicit with goals and intents, organizations must make sure their data hygiene is sound, Kinha says.

The future looks bright(ish) — but unpredictable

When the technical and process challenges are reconciled, HR and IT partnership will be essential in assisting the transition from humans to human-plus-machine work. Every company introducing AI agents to their organizations must become more intentional about how they execute their business processes and measure outcomes.

“All of us in different functional domains need to up our game in intent-setting, boundary-setting, and measurement,” Hjartar says. “That’s going to take many years for us.”

Young says that every company will proceed at their own pace, which will create new categories of haves and have nots — just like preceding paradigm shifts involving emerging technology. “Some will push hard to automate; others won’t.”

What’s clear is that the challenges of human-machine commingling in the workplace are just beginning.