Agentic artificial intelligence (AI) is being hailed as the next major leap in enterprise technology, promising to transform industries by 2028. A new report from Capgemini Research Institute estimates the technology could generate up to US$450 billion in economic value within the next three years. Yet despite the potential, adoption remains slow: only 2% of organisations have scaled its use, reflecting both scepticism and uncertainty about its role in business.
Staff Writer
Trust, oversight, and the human factor
The Capgemini study, based on a survey of 1,500 executives across 14 countries including Singapore, points to a central theme: agentic AI works best when paired with human oversight. Nearly three-quarters of executives said the benefits of human involvement outweigh costs, and nine out of ten agreed that oversight adds value or at least does not add significant expense.
In practice, this means companies see AI agents not as standalone replacements but as collaborative tools—ones that can automate tasks while still requiring human judgment to ensure accountability.
Early steps, uneven progress
While about a quarter of organisations have launched pilot programmes, only 14% have moved into implementation. For most, deployment remains stuck at the planning stage, widening the gap between intent and readiness.
Still, real-world use cases are emerging. Retailers are testing AI-powered shopping assistants that can search for products, generate descriptions, and manage shopping carts through voice or text. Though these systems typically stop short of completing transactions for security reasons, they mimic many functions of a human sales assistant and raise new questions about the future of online shopping.
What makes agentic AI different?
Unlike generative AI, which reacts to prompts and creates content, agentic AI is designed to act independently, pursue objectives, and adapt strategies in real time. Jason Hardy, Chief Technology Officer for AI at Hitachi Vantara, described it as “a team of domain experts that can learn from experience, coordinate tasks, and operate in dynamic environments.”
This distinction—between producing outputs and driving outcomes—helps explain why enterprises are paying attention.
Why enterprises are adopting it
According to Hardy, the momentum behind agentic AI comes from scale and complexity. Enterprises today face mounting challenges in IT operations, compliance, and security. Agentic AI promises not just insights, but autonomous action—optimising storage, anticipating failures, automating governance, and responding to threats in real time.
Capgemini’s research backs this up, showing that initial deployments are most effective when the technology takes on routine but critical IT functions. Early users are already seeing benefits in areas like predictive maintenance, cybersecurity response, and compliance reporting.
Southeast Asia’s priorities
For Southeast Asia, readiness starts with data quality and infrastructure. Hardy stressed that enterprises need properly classified, secured, and governed data before agentic AI can deliver meaningful value. Robust systems capable of supporting multi-agent orchestration and dynamic resource allocation are also essential.
IT remains the most practical entry point, particularly for organisations looking to improve performance and reduce downtime. From there, businesses may expand to supply chain management and customer service, where agentic AI could reshape workflows and improve resilience.
Skills, jobs, and leadership
Agentic AI will shift the role of people from executing tasks to orchestrating and overseeing AI systems. This transition requires new skills in governance, auditing, and ethical oversight.
The World Economic Forum projects that by 2030, AI could create 11 million jobs in Southeast Asia but displace about nine million, with women and younger workers facing the highest risk of disruption. Investments are already underway: Microsoft has committed US$1.7 billion in Indonesia and is expanding training programmes in Malaysia and across the region to support workforce reskilling.
Looking ahead
IDC forecasts that AI, including generative and agentic systems, could add around US$120 billion to ASEAN-6 GDP by 2027. Hardy believes the transformation may unfold faster than leaders expect, with IT operations already showing measurable results.
Deepika Giri
The challenge, however, is striking the right balance. As Capgemini’s research and Hardy’s insights underline, the real promise of agentic AI lies in combining autonomy with oversight. Companies that build strong governance, invest in workforce readiness, and modernise infrastructure will be best positioned to capture its economic and operational benefits.
For Southeast Asia, the question is no longer whether agentic AI will take root, but how quickly businesses can scale it responsibly.
