Across the sprawling archipelagos of Indonesia and the Philippines, a massive economic engine remains stalled — not because entrepreneurs lack hustle, but because they lack legibility in the eyes of banks.
Millions of micro, small, and medium enterprises (MSMEs) are effectively invisible to traditional lending systems. Without formal credit histories, audited statements, or pristine collateral, these businesses are routinely excluded from the capital they need to scale. This is not only a social issue but a commercial blind spot at regional scale.
Also Read: Southeast Asia’s banks have entered the AI revenue era
The executive insights report, titled “From Pilots to Production: How Banks Turn AI into Revenue” by Dyna.AI, GXS Partners, and Smartkarma, estimates the ASEAN MSME finance gap at a staggering US$300 billion.
For years, the bottleneck has been credit scoring itself: rigid models that privilege historical bureau data and formal documentation — precisely what many warung owners, sari-sari store operators, market traders, and home-based sellers do not have.
AI is now changing that equation by shifting the centre of gravity from risk exclusion to risk pricing and ultimately towards financial inclusion that is profitable rather than philanthropic.
Alternative data turns “thin-file” into under-writeable
The region’s digital behaviour footprint has quietly become one of ASEAN’s most valuable assets. By leveraging alternative data (including telco records, e-commerce transaction histories, mobile wallet usage, point-of-sale flows, logistics and delivery patterns, and even psychometric indicators), AI models can underwrite “thin-file” borrowers with far more precision than traditional scorecards.
In Southeast Asia, this is especially powerful because MSMEs increasingly operate digitally even when they are informal. A seller may not have an audited P&L, but they may have:
- A year of daily transactions on Shopee, Lazada, Tokopedia, or TikTok Shop
- Repeat-customer behaviour visible through e-wallet and QR payments
- Inventory turnover patterns in POS systems used by small retailers
- Repayment signals from buy-now-pay-later (BNPL) or supplier credit
- Telco top-up regularity and geolocation stability patterns (where permitted).
The report notes that effective AI-driven personalisation in lending can lift revenues by 10 to 25 per cent. For a mid-sized regional bank, capturing even a slice of the underbanked MSME segment can translate into hundreds of millions of dollars in additional income — not just from interest, but from deposits, payments, insurance, and merchant services.
Philippines: From static lending to dynamic assessment
In the Philippines, where a significant portion of the population remains underbanked, lenders have been leaning into partnerships with AI-as-a-service providers and fintech infrastructure players to modernise decisioning while keeping portfolio quality steady. The direction of travel is clear: lenders want models that update risk views continuously instead of freezing them at the moment a form is signed.
That “dynamic” approach matters in an economy where income can be seasonal, informal, or platform-linked. For example, a micro-merchant’s risk profile can improve sharply as their digital sales stabilise, their returns drop, and their fulfilment performance improves– signals that static models often miss.
This is also where the Philippines’s strong remittance and mobile-money culture becomes relevant. Regular inflows, bill payment behaviour, and wallet velocity can form a proxy for stability when traditional documents are absent.
These models do not just guess; they use predictive analytics to forecast behaviour and risk based on real-time data signals.
Indonesia: QRIS data makes MSMEs visible at scale
Indonesia provides another compelling case study for revenue-generating inclusion, and it comes with a national data exhaust pipe: payments.
Also Read: Bridging the credit gap: CBI’s new bureau targets MSME financing bottlenecks
The rapid adoption of the QRIS national payments network has created a treasure trove of behavioural data. With 39.3 million merchants (93 per cent of them are MSMEs) connected to the system, transaction volumes have surged by 175 per cent year-on-year. Banks and lenders are now deploying AI to identify high-potential merchants within this ecosystem, automating onboarding and offering credit lines based on digital cash flow rather than collateral.
The implications are huge. Once a merchant’s daily QRIS sales can be tracked, lenders can structure products that match reality:
- Revenue-based repayment tied to daily receipts
- Short-tenure working capital for inventory cycles
- Pre-approved limits that expand as payment consistency improves
- Faster renewals with fewer manual checks
This is how lending becomes a scaled growth engine: distribution via embedded channels, underwriting via data, and servicing via automation.
Beyond Indonesia and the Philippines: ASEAN’s rails are converging
While Indonesia and the Philippines are headline examples, similar dynamics are playing out across the region:
- Thailand’s PromptPay and QR adoption have normalised low-friction digital payments for small merchants, improving cash-flow visibility.
- Malaysia’s DuitNow and the broader push for digital payments give lenders more structured signals for MSME activity.
- Singapore’s PayNow and the city-state’s dense SME ecosystem create a testing ground for model governance, though scale often lies north and east, where informality is higher.
Even smaller markets are building rails that generate the data needed for AI underwriting. Cambodia’s Bakong system, for instance, has helped accelerate digital payments adoption, which can support more data-led credit products over time.
The commercial transformation and the caution
As the white-paper notes, “AI shifts lending from exclusion to inclusion”. That shift is commercially transformative because it converts untapped customer segments into profitable borrowers — customers traditional models could not touch.
Global comparisons reinforce the point. In Latin America, similar AI-based credit scoring has been shown to outperform conventional models by up to 85 per cent in accuracy. ASEAN’s digital platforms are just as data-rich; the constraint is less “data existence” and more “data usability”.
The hardcore challenge lies in infrastructure and governance:
- Regulatory fragmentation: An AI model validated in Singapore or Malaysia often needs significant tailoring before it can be deployed in Indonesia, Vietnam, or the Philippines, given differences in data privacy, model risk management expectations, and permissible data sources.
- Consent and trust: Alternative data can expand inclusion, but only if customers understand what is being used and why — and if regulators are comfortable that models are fair, explainable, and auditable.
- Talent scarcity: The region still lacks enough professionals who understand both AI and the nuances of local financial regulation, credit risk, and consumer protection.
Why the momentum is still undeniable
Despite those hurdles, the flywheel is turning. With more than US$30 billion recently committed to AI-ready data centre infrastructure in Singapore, Thailand, and Malaysia, the foundation for scalable AI deployment is being laid — not only for consumer use cases, but for the heavy lifting required in lending: feature stores, real-time decisioning, monitoring, and compliance tooling.
Also Read: How Ant International bridges MSME finance gap with intelligent credit services in the AI era
The winners in this next phase will be the institutions that move fastest from “interesting” pilots to production-grade lending models: turning ASEAN’s “invisible” entrepreneurs into a compounding source of revenue, while expanding access to capital in the places that need it most.
—
The image was generated using AI.
The post From invisible to investable: How AI is unlocking ASEAN’s MSME goldmine appeared first on e27.