When generative AI first captured global attention in late 2022, the appeal was straightforward: here was a technology that could create. It could write essays, draft emails, generate images and mimic tone in ways earlier automation tools never could. It felt new, expressive and accessible, like something anyone could try.
Back then, generative AI was used for small everyday tasks like summarising meeting notes, to even helping teams brainstorm ideas. These early interactions showed how naturally it could fit into workflows where speed and clarity mattered, prompting organisations to rethink how work could be streamlined. The clearest opportunity was in functions that dealt with high volume, repetitive interactions, especially customer-facing teams. This is where the shift from simple automation to genuinely intelligent support began.
Here is a GenAI toolkit for SMEs in Southeast Asia
In Southeast Asia, the momentum is now obvious. Although only 23 % of organisations in the region are classed as “transformative” in their AI adoption, the economic stakes are high: research estimates that AI and generative AI could contribute up to US$120 billion to Southeast Asia’s GDP by 2027.
From chatbots to AI agents
Prior to generative AI, most companies relied on rule-based chatbots that matched keywords to preset answers, which was helpful for basic FAQs, but often frustrating when questions fell outside the script. Most customers learned to recognise the pattern: ask anything slightly unusual and the chatbot would escalate to a human agent, but this didn’t solve the problem.
Generative AI changed things by introducing context and language understanding. Instead of searching for exact phrases, the AI could grasp the intent of a request, even if it was phrased casually, emotionally or with mixed languages, which is extremely common in Southeast Asian countries where most people speak multiple languages.
The next step was even more impactful: generative AI powered chatbots didn’t just select an answer. It could draft a response which was clearer, warmer and even aligned to the company tone. Support agents remained in control, but the mental load of writing the same explanations repeatedly began to ease.
And we are now seeing the shift to AI agents that can actually perform tasks. Instead of simply responding to customer inquiries, AI systems are being connected directly to business platforms such as logistics systems, billing systems, CRM databases and booking platforms, allowing them to:
- Check or update order status
- Process refunds or replacements
- Schedule deliveries or appointments
- Create and close service tickets
- Surface customer history instantly
From there, adoption expanded. Marketing teams began using generative AI to create campaigns with localised messaging and HR teams leaned on it to rewrite job descriptions and standardise onboarding materials. These weren’t headline-stealing transformations, but they revealed something important: generative AI wasn’t just a creative novelty but was becoming a practical productivity layer across everyday work.
As the technology matured, its role widened again. Organisations started applying generative AI to document processing, multilingual translation and intelligent customer engagement. This has been especially meaningful in Southeast Asia, where businesses operate across multiple languages like Malay, Mandarin, Thai, Tamil, Tagalog and English. And more often than not on the same day!
Now, after two full cycles of hype and experimentation, generative AI is moving into its pragmatic enterprise phase. In 2026, the real question won’t be what AI is capable of but how it will be effectively integrated, governed and scaled to deliver measurable improvements in cost efficiency, productivity and customer experience.
The real era of Gen AI begins
Across the region, early adopters are already demonstrating what this looks like in practice. In Singapore, DBS Bank reported a 20% reduction in average handling time after deploying generative AI to support frontline service agents, not by replacing them, but by drafting responses, retrieving case histories and suggesting next steps in real time. In Malaysia, CelcomDigi’s Maya chatbot demonstrates how generative AI can take on a cultural role as well as a functional one. Maya is designed to help users explore and interact with their ancestral languages, showing how AI can strengthen connection to identity while improving accessibility for multilingual communities. These deployments are practical, localised and context-aware and that is precisely why they work.
Taken together, these examples signal the direction Southeast Asian enterprises are heading as we approach 2026. The competitive advantage will not come from being the first to experiment with AI, but from being able to operationalise it consistently, with the right guardrails, multilingual adaptation, data controls and workflow integration.
Generative AI will not replace teams but it will reshape how they allocate time, attention and expertise. The organisations that succeed will be those that align the technology with real business needs, measurable outcomes and governance frameworks that build trust across both customers and employees. This is the phase where generative AI stops being a trend and becomes infrastructure.