Southeast Asia is counting what it automated, it is not counting what came back

Southeast Asia is not lagging behind in AI adoption, according to published statistics. In fact, it appears to be ahead of the global average. However, the true challenge lies beneath the surface. A McKinsey survey of 2,000 firms in the region found that 46% have moved past pilots, compared to only 35% globally. Yet, nearly 80% of these companies report no significant impact on their bottom line.

The issue goes beyond shallow adoption; there's an overlooked consequence that no survey measures. When a company hands work to a machine, it also takes something back. Companies typically record what they automated, but not what they had to return to human workers. In this story, two 23-year-old co-founders built a company to uncover where the machine stops and what comes back.

They ran the model in three Southeast Asian markets – Singapore, Vietnam, and Malaysia. The key insight is that the amount of work left behind depends on how much of it is already documented. When processes are clear, recorded, and priced, more of it stays automated. Conversely, when processes rely on relationships and undocumented judgment, more comes back to humans.

This isn't a maturity ranking but rather a reflection of the existing documentation level. Countries that fail to recognize this will struggle in the coming years, as they focused on counting their AI investments rather than understanding what they lost by automating. The crucial question is: what did you take back? An organization that cannot answer this question has not truly tested the AI implementation.