The Asia Pacific region offers unique opportunities in scaling agentic AI. However, it also presents distinct challenges that organisations must navigate to scale the technology successfully.
Shashir Shetty, APAC Google Cloud GTM leader at NTT DATA, shared the region’s unique challenges in scaling agentic AI and how organisations can overcome them.
APAC’s unique challenges
According to Shetty, APAC’s unique challenges stem from its regulatory diversity, infrastructure and market maturity.
He said: “Unlike traditional AI, agentic AI introduces additional complexity around governance, autonomy and accountability, which are harder to standardise across multiple jurisdictions.”
One major hurdle, according to Shetty, is APAC’s fragmented regulatory landscape. Unlike regions such as the European Union, where regulatory frameworks are more harmonised, organisations in the Asia Pacific must navigate different data residency laws, AI governance frameworks and compliance requirements in every market.
Beyond regulation, Shetty said that data sovereignty and infrastructure constraints are also major considerations. He cited NTT DATA’s research that revealed that an overwhelming majority (94%) of organisations in Asia Pacific consider private or sovereign AI important to their AI strategy, while 93% are considering relocating AI infrastructure due to geopolitical and supply chain concerns.
“This highlights how data control, sovereignty and resilience are becoming central to AI decision-making,” he explained.
At the same time, multiple jurisdictions in the APAC region have different regulatory expectations.
“Unlike the European Union, where sovereign AI investment is more regulation-driven, and parts of the Middle East, where national strategy often plays a larger role, APAC requires balancing innovation, governance and compliance across significantly more diverse markets,” Shetty said.
He added: “Overall, APAC’s challenge is not just scaling AI but doing so across a highly heterogeneous environment while balancing innovation, compliance and control.”
Overall, APAC’s challenge is not just scaling AI but doing so across a highly heterogeneous environment while balancing innovation, compliance and control. Shashir Shetty
Correcting misconceptions
Beyond external challenges, organisations also need to overcome misconceptions about agentic AI.
For Shetty, one of the biggest misconceptions is that scaling agentic AI is primarily a model challenge.
“Most organisations already have access to powerful AI models. The bigger challenge is operationalising AI across the enterprise,” he said.
He explained that underestimating the importance of infrastructure, governance, data readiness and process integration leads to bottlenecks that prevent organisations from scaling agentic AI successfully.
NTT DATA research supports this, revealing that almost all (96%) organisations in Asia Pacific believe legacy infrastructure is slowing AI adoption.
Another misconception, according to Shetty, is that AI agents can operate independently.
“While agentic AI can automate increasingly complex tasks, trust remains non-negotiable,” he said.
He reminded organisations to maintain human accountability, governance and oversight in scaling agentic AI.
“The source of the greatest competitive advantage is not the model alone, but the ability to deploy AI securely, responsibly and at scale. Competitive advantage comes not from access to AI models, but from the ability to deploy them securely, responsibly and at scale,” he said.
Overcoming challenges
Once misconceptions are corrected, Shetty believes organisations can focus on what it truly takes to scale agentic AI successfully.
“Leading organisations are shifting from experimentation to operationalisation, recognising that AI itself is only a small part of the equation. As reflected in the “1-2-3-4 rule”, for every $1 spent on AI technology, organisations need to invest $2 in adoption, $3 in building the ecosystem and $4 in preparing data,” the NTT DATA executive explained, highlighting data readiness and organisational enablement.
He urged CIOs to build strong data and infrastructure foundations that align AI strategy with data architecture, platforms and security.
“Without high-quality, well-governed data, agentic AI systems cannot operate reliably at scale,” Shetty said.
Equally important is governance. As AI agents become more autonomous, prioritising accountability, risk management processes and policies for responsible AI use will help sustain trust.
Shetty also stressed the need to drive enterprise-wide adoption and integration. Rather than treating AI agents as standalone tools, CIOs should embed them into business workflows, redesign processes and ensure interoperability with legacy systems.
Lastly, workforce readiness should not be overlooked.Investing in employee training will help people work alongside AI systems.
For Shetty, AI adoption is more than a technology, governance, and compliance issue; it also involves people. He believes that Organisation Change Management (OCM) has a vital role in sustaining AI adoption at scale.
“Ultimately, scaling agentic AI requires a coordinated approach across technology, governance and people, ensuring these elements evolve together to deliver measurable business outcomes,” Shetty shared.
Helpful metrics
As organisations scale agentic AI, measuring success should go beyond technical performance.
“This starts with business outcome metrics, including productivity gains (e.g. time saved per task), operational efficiency (e.g. cost reduction or throughput improvements), customer experience and the extent to which workflows have been redesigned rather than simply automated,” Shetty shared.
He added that organisations should also track operational and decision-making metrics when it comes to agentic AI. These indicators measure task completion rates, outcome accuracy, level of autonomy (versus human intervention), and escalation or error rates; they help assess whether AI agents are operating reliably and delivering consistent results.
“Governance and trust should also be part of the measurement framework,” Shetty reminds organisations. Embedding AI into core operations must move organisations to assess security, compliance and governance structures.
Lastly, Shetty believes that adoption is a key indicator of success.
“AI transformation happens when employees can effectively integrate AI into daily workflows, with humans remaining accountable for critical judgment and business outcomes,” he said.
Scaling agentic AI
“If I could give CIOs in APAC one piece of advice, it would be this: focus on operationalising AI, not just deploying it,” Shetty shared.
He believes that a mindset shift is what matters most.
“Advantage does not come from choosing the right model; it comes from the foundations underneath — data that’s clean and accessible, infrastructure designed for continuous AI workloads, governance embedded from the start, and an ecosystem of partners that strengthens what you can deliver. Adoption is where transformation happens,” he shared.
Your competitive advantage will not come from choosing the right model, but from the ability to embed AI into workflows at scale, with strong governance and human accountability. Shashir Shetty
He concluded: “Your competitive advantage will not come from choosing the right model, but from the ability to embed AI into workflows at scale, with strong governance and human accountability. That is what turns experimentation into sustained business value.”
Unique challenges, unique opportunities
Scaling agentic AI in APAC comes with unique challenges.
However, organisations that succeed will be those that view these challenges not as barriers, but as opportunities to build stronger AI foundations.
In a region as diverse as Asia Pacific, the ability to balance innovation, governance and people will ultimately determine who scales agentic AI successfully.
