Orchestration: The new OS for Southeast Asian enterprises

As Southeast Asian (SEA) organisations accelerate AI adoption, many are grappling with scaling use cases without compromising security and trust. Concurrently, with more SEA companies deploying agentic AI, we are starting to see a reallocation of work where AI agents handle complex decisions and workflows, robots manage repetitive tasks, and humans take on strategic roles.

A recent IDC InfoBrief, Agentic Automation: Unlocking Seamless Orchestration for the Modern Enterprise, commissioned by UiPath, shows strong momentum behind the use of AI agents for businesses in SEA, but has also highlighted key gaps in oversight and orchestration of AI agents, robots and humans in the workforce.

Matthew Tan, Principal Solution Engineer at UiPath, Southeast Asia (MT), shares how businesses can navigate the challenges to establish strong governance frameworks for responsible agentic AI and automation, rethink collaboration, and redesign job roles.

Q: What are some unique challenges and opportunities in agentic AI adoption in SEA?

Source: UiPath. Tan.

MT: Agentic AI holds great promise in SEA. In Singapore alone, AI technology spending is projected to grow yearly by 30%, reaching US$7.8 B by 2028. As digital transformation accelerates, enterprises have the opportunity to improve efficiency, scale operations, and deliver more personalised service experiences by adopting agentic AI.

The new IDC InfoBrief revealed a growing adoption of AI agents by SEA organisations, with 86% of them expected to use AI agents within the next 12 months. Four in five (79%) SEA organisations are actively developing use cases for agentic AI, even if they have not yet made substantial investments. These findings reflect a broader trend of organisations shifting from AI experimentation to large-scale implementation.

However, trust and security remain major hurdles with agentic AI adoption. SEA organisations are concerned with business risks associated with agentic AI, primarily data privacy breaches, security vulnerabilities from autonomous actions, and unintended consequences from complex interactions.

SEA countries also face challenges that affect the adoption of agentic AI, including a wide gap in digital maturity due to varying infrastructure, and differing regulatory frameworks and data governance rules. 

How can companies use agentic AI effectively to achieve business outcomes? 

MT: To unlock the full value of agentic AI, enterprises must use agentic automation to orchestrate AI agents, robots, and human teams responsibly, securely, and effectively. Agentic automation integrates agentic AI and robotic process automation (RPA) into workflows, unlocking efficiency, scalability, and innovation while having humans in the loop. Solutions like the UiPath Platform enable companies to orchestrate a workforce where AI agents think, robots execute, and humans stay in control to drive smarter decisions and more business value.

On the implementation front, companies should ensure that AI agents are built to achieve business goals. They can start by identifying key business challenges where AI agents can drive clear ROI such as reducing operational bottlenecks or improving customer responsiveness. Leaders should also foster transparent human-agent collaboration with robust governance on scalable platforms and upskill their workforce.

How has agentic AI empowered employees?

MT: Agentic AI systems are designed to support human work by removing repetitive and rules-based tasks, and SEA organisations are beginning to see clear benefits. According to the IDC InfoBrief, 75% of respondents say agentic AI improves decision-making, while 72% report increased productivity, suggesting that employees are better equipped to act on insights, adapt to change, and contribute at a more strategic level.

In customer service, for instance, AI agents can assist by triaging requests, retrieving information, and recommending next steps. This allows employees to focus on customer engagements that are more complex or sensitive, which requires human judgment and experience. In healthcare, AI agents can analyse medical data, suggest potential diagnoses, and recommend personalised treatments. Yet, the final decision still rests with the doctor, who now has more time to engage with patients and provide them with better care.

With AI agents taking on more complex decisions, what should organisations focus on? 

MT: As AI agents become more autonomous and make increasingly complex decisions, organisations must strengthen their human oversight and governance structures to maintain trust and accountability. This means embedding clear ethical guidelines that define acceptable AI behaviours, ensuring transparency in AI decision-making processes, and implementing robust monitoring systems that allow humans to review and intervene when necessary.

A good practice is to have a dedicated trust layer with strong access controls, real-time analytics, and detailed reports on how AI is being used across business processes. A human-in-the-loop approach is also essential, where humans retain ultimate control over critical decisions and can override AI actions when needed. Additionally, organisations should establish cross-disciplinary ethics committees and regularly audit AI systems to safeguard against bias, errors, and unintended consequences.

How might the roles of AI agents, robots, and humans evolve in SEA workplaces?

MT: As we enter the agentic era, we will see a reallocation of work across AI agents, robots, and humans, in SEA. Each will play a distinct role: 

- AI agents will bring intelligence and autonomy to manage more complex workflows and decision-making; 

- Traditional robots will automate rule-based, repetitive tasks; and 

- Humans will focus on what they do best — strategic thinking, empathy, and creative problem-solving. Humans will also “monitor the loop” to provide stronger governance and orchestration, stepping in only when judgment is needed.

Picture a manufacturing company whose purchase-to-pay (P2P) process is streamlined through agentic automation. When a requisition is raised, an AI agent checks inventory levels and budget alignment, then a robot generates and sends a purchase order (PO). 

Once the vendor invoice arrives, another AI agent extracts and matches the data against the PO and goods receipt. If everything aligns, a robot posts it to the enterprise resource planning (ERP) software and triggers payment. For discrepancies, the AI routes the case to a human with context and resolution suggestions.

In this orchestrated model, humans remain essential — making final decisions, managing exceptions, and bringing context and domain expertise that technology cannot replicate. The seamless collaboration between AI agents, robots, and humans improves efficiency and scalability while enabling more human- centred innovation across the enterprise.

*OS stands for operating system.