Southeast Asia’s SMEs do not have an AI problem. They have a record problem

Core Argument

The piece opens on a Colombo distributor proudly demonstrating a demand forecasting dashboard. Clean charts, confidence bands, reorder quantities. One question dismantles it: the underlying order data came from a junior staffer retyping WhatsApp messages each evening, sometimes guessing when a field rep wrote "the usual." The model was competent. It was forecasting a fiction at speed, with a confidence interval attached. From there the article rejects the region's dominant framing. Southeast Asia's SME conversation has settled on adoption, meaning which tools, what budget, how fast. That is the wrong variable. AI does not create operational intelligence. It multiplies whatever operating record a business already keeps. Most SMEs in the region are multiplying by close to zero. The argument then moves across scale. For the owner-operator, AI is a language tool and nothing more. The break happens between twenty and two hundred staff, where a business has enough volume to need a system and not enough discipline to own one, leaving three competing versions of the truth. At regional operator scale the record fragments across currencies, tax regimes and languages. Only at enterprise scale does the problem invert into governance rather than capture. Two concrete traps follow. First, the real cost of AI is reconciliation labour, which lands on payroll rather than the software line and therefore never appears in the business case. Second, off-the-shelf forecasting assumes fixed-date seasonality, while Ramadan, Tet, Lunar New Year and Vesak all move, guaranteeing a mis-forecast of the year's largest sales window. The macro frame: MSMEs are 97 to 99 percent of ASEAN enterprises and roughly 85 percent of employment, but near 40 percent of GDP. That gap is a systems gap. The close lands the reframe. The businesses that scale on AI here will not be the ones that bought first, but the ones that wrote it down first.