Leading businesses in Thailand have begun to embrace agentic artificial intelligence (AI) to pilot selective tasks, in a bid to explore its benefits.
The term refers to an AI system with enhanced autonomy, decision-making capabilities and adaptability. Agentic AI systems autonomously plan and take actions to meet user-defined goals, offering the promise of a virtual workforce that can offload and augment human work.
According to global research and consultancy Gartner, agentic AI is a top strategic technology trend in 2025 that is poised to drive innovation, offering significant business benefits by providing a virtual workforce of AI agents.
However, the implementation of agentic AI also presents challenges, particularly in ensuring that these autonomous systems are aligned with the intentions of both the providers and users.
To mitigate these risks, it is crucial to establish robust safeguards and guardrails that guarantee AI's actions remain consistent with the desired outcomes, said Gartner.
The company predicts by 2028, 33% of enterprise software applications will include agentic AI, up from less than 1% in 2024, enabling 15% of day-to-day work decisions to be made autonomously.
First in the region
"In 2025, agentic AI will be a key trend, combining AI and automation," said Anothai Wettayakorn, managing director of IBM Thailand.
The rise of agentic AI marks a significant shift in the future of work, with AI agents autonomously executing tasks, collaborating with human workers, and driving value across businesses, he said.
In Thailand, IBM successfully completed a pilot initiative using generative AI (GenAI) to automate laboratory orders, enhancing patient diagnosis and improving confidence, quality and speed in treatment.
This collaboration took place with Maharaj Nakorn Chiang Mai Hospital, an affiliate of Chiang Mai University's Faculty of Medicine.
This was the first agentic AI use in Southeast Asia for healthcare, said Mr Anothai.
After an eight-month pilot and an investment of 10 million baht, IBM's agentic AI technology leverages existing automation in the hospital's laboratory and hospital data system to speed up patient services.
The initiative minimises repetitive tasks for doctors and nurses, said Dr Bannakij Lojanapiwat, faculty dean at the university.
IBM's client engineering team worked alongside the faculty to build a model using IBM AI platforms to analyse and interpret doctors' notes, recommend treatment pathways, and automatically handle laboratory orders.
AI can suggest pathology laboratory tests based on a patient's medical record and trends, and doctors can then verify the suggestions.
Once approved by doctors, the AI automates the lab orders and schedules the patient's follow-up appointments, alleviating the workload for both doctors and nurses, said Dr Bannakij.
The collaboration complemented the hospital's recent implementation of a fully automated medical laboratory system.
Built on transparent, reliable data and global AI governance standards, this initiative was meant to improve service efficiency and support medical decision-making with accurate and comprehensive insights, enhancing treatment quality and speed, according to IBM.
The project eases high-volume workloads, reducing the time spent in the lab order process and on scheduling patient appointments. As a result, it cuts the time spent by patients in medical check-ups at the hospital by 30-40 minutes, down from the current 150 minutes, he said.
The hospital has around 1.6 million patients a year. For the project, physicians also help to train the AI system.
"This milestone highlights our vision to lead in service excellence, treatment and research -- not just in Chiang Mai or Thailand, but across Southeast Asia," said Dr Bannakij.
Dr Krit Khwanngern, deputy dean of the Faculty of Medicine at Chiang Mai University, said the system does not store patients' real names, instead anonymising them to protect sensitive data and comply with the Personal Data Protection Act.
In the future, the hospital will analyse patient medical check-up data and use it to offer preventive healthcare services on a more personalised basis, said Dr Krit.
"Agentic AI is leveraged with the digital improvements we've invested in over the past few years by integrating different systems from physicians, nurses, pharmacies, and electronic medical records for both inpatient and outpatient care. This provides a strong data infrastructure on the back end, which leads to improved services on the front end," he said.
"We also invested 300 million baht in cybersecurity to ensure strong capabilities to secure sensitive data."
The project was successful because of the ability of the hospital's in-house teams to work together with external tech experts, while its physicians have technical knowledge that can design systems to serve users' needs, said Dr Krit.
The hospital continues to adopt advanced technologies, he said.
The faculty piloted IBM's AutoAI to support the development of patient service apps, enabling data scientists to write code more efficiently and reduce development time for future apps that will enhance patient treatment and services, said Dr Krit.
Many executives note that humans will still be required to operate AI in a responsible manner.
Agentic AI in banking
"2025 will be the year of agentic AI and we've been working on responsible agentic AI since the beginning of this year," said Ruangroj Poonpol, chairman of Kasikorn Business-Technology Group (KBTG), the technology arm of Kasikornbank.
Agentic AI will be an essential tool for KBTG to accelerate its organisational transformation, he said. An AI agent can perceive, make decisions, take action and achieve goals in its own environment.
"It's a digital workforce without the need of a vacation, providing even more productivity uplift than before. We need to upskill, reskill, and learn how to use it and when to use it," said Mr Ruangroj.
In 2025, KBTG will focus on fostering collaboration between humans and AI, while ensuring a human-centric approach, he said. With human-AI convergence for value creation, KBTG's performance will be faster and more productive than ever, said Mr Ruangroj.
He said this focus will be driven by a few elements, including the use of new platforms to efficiently manage AI agents, enabling seamless collaboration with human workers.
One of the platforms is AthenaMind, an internal tool designed as an AI agent creation hub, or "AI agent factory", for building AI agents for various applications.
Another element is an agentic orchestration, involving the design of new operational workflows to enable AI agents and humans to work together more effectively and drive business outcomes.
For example, KBTG introduced an AI coding assistant to improve software development, reducing software developers' working hours by 20%. So far, more than 500,000 lines of code have been generated using AI.
The next step is to evolve this AI assistant into a fully functional AI coding agent, capable of independently writing code from start to finish based on clear requirements, said Mr Ruangroj.
KBTG equips its employees with the right mindset, skills, and an open work culture to integrate AI into their daily tasks, he said. The company achieved 100% AI literacy, ensuring all employees are knowledgeable and prepared to work alongside AI technology in their roles.
KBTG wants this human-AI convergence project to generate more than 10 billion baht in business impact by 2029, said Mr Ruangroj.
Retail uses
Axel Winter, chief executive of Xponential, a software-as-a-service (SaaS) platform and innovation consultancy, said the opportunities for AI agents are vast in the retail sector.
"For example, agents could dynamically create promotions based on real-time data such as customer sentiment, inventory levels and sales trends," he said.
They can also generate and deploy marketing campaigns across various channels, which are particularly relevant in the fast-paced Southeast Asian e-commerce landscape.
"It's important to recognise not all AI technologies being promoted are fully mature, and the field is in constant flux. It is vital to remain flexible, keep a close eye on both established vendors and the open-source community, and be prepared to adapt and innovate rapidly," said Mr Winter.
Widespread in 2026
Phiphat Prapapanpong, director of advanced insights at Bluebik Group, a SET-listed tech consulting firm, said the adoption of AI-driven agentic workflows, which leverage GenAI capabilities with business processes, will emerge as a key trend in AI development.
"This year marks the beginning of the agentic AI era, with large enterprises -- particularly banks, insurance companies and retailers -- starting to pilot its use internally," he said.
By next year, Mr Phiphat said he expects agentic AI to become more widespread and transition into full deployment.
He said organisations piloting agentic AI aim to enhance employee productivity without additional hiring. For example, procurement processes can use agentic AI to seek better prices and sourcing.
Job losses and opportunities
Panutat Tejasen, a pioneer in AI development, said the adoption of agentic AI is accelerating because it can integrate with business workflows, delivering tangible results for return on investment.
He said many jobs will be replaced as agentic AI takes over tasks. For example, roles in customer service, procurement and data entry are likely to be automated.
"You will see machine agents communicating with other machines. A procurement agent, for example, will negotiate with a sales agent, with less human involvement," said Mr Panutat.
He said agentic AI will also open up new opportunities in software development.
Thanachart Numnonda, executive director of IMC Institute, a tech research and training firm, said agentic AI is still being used for specific and small-scale tasks.
"This opens up new opportunities for developers to build agentic AI services tailored for businesses," he said.
Mr Thanachart said he does not think agentic AI will make the SaaS business obsolete, as it still requires software as its foundation.
"Humans are still in the loop and work alongside AI," he said.
