Embedded Finance, AI Agents, and Stablecoins: The Next Wave for Southeast Asia's Digital Economy
A new regional study from HSBC, Google Cloud, and Payments and Commerce Market Intelligence surveyed 2,400+ respondents across six major Southeast Asian markets between December 2024 and May 2025. The data points to three forces moving in tandem: embedded finance adoption, agentic AI, and real-time value transfer powered by instant rails and stablecoins. If you work in finance, this isn't abstract. It's a product roadmap, a P&L lever, and a risk checklist.
Embedded finance is now default
Embedded finance has crossed the threshold from nice-to-have to baseline. 77% of consumers in Southeast Asia use wallets, BNPL, or in-app loans, and 75% call these features very important or useful. That's behavioral lock-in.
Consumers want finance where they already spend time. Integration with social platforms ranks as the number two feature that would improve their digital banking experience. One in three want financial services in messaging apps, and one in four want them in gig platforms and other daily-use apps.
Embedded credit demand is rising-and it skews premium
Demand clusters around credit at the point of need. 66% want to see BNPL more often and 48% want instant loans. The surprise: 61% of BNPL users sit in the top two income quartiles, with 69% holding credit cards versus 45% on average, and 75% transacting digitally every day.
Translation: BNPL is a discretionary spend optimizer for higher-income, digital-first users. Expect strong pull in luxury, travel, lifestyle, and high-end electronics. Finance teams should price for lower loss rates than commonly assumed in BNPL and design premium tiers that bundle perks, insurance, and flexible repayment.
Agentic commerce is getting real
Agentic commerce-autonomous AI agents acting on behalf of customers or businesses-is moving from concept to rollout. 73% of respondents trust AI with their personal data, or at least don't mistrust it. Regulators are laying foundations; for example, Singapore's Cyber Security Agency issued practical guidance on securing agentic AI systems.
Industry intent is visible. Research indicates a meaningful share of organizations plan deployments by 2026, and many expect business model disruption on a short timeline. Pair those agents with programmable money, and you get always-on commerce flows where bots not only decide but also settle transactions under preset rules on amount, merchant type, or payment method.
What to do now:
- Stand up an "agent-safe" product surface: scoped permissions, consent flows, and per-action limits.
- Instrument clear audit trails and policy controls for autonomous decisions.
- Pilot programmable payments (including stablecoin rails) with strict transaction policies and thresholds.
AI personalization is the expectation
Personalization started in e-commerce and is now standard in finance: 88% are accustomed to it, and 84% would switch to a bank or payment provider that offers more personalized services. The features with the most pull: budgeting intelligence (53%), investment advice (48%), and spending insights (43%).
Action for product owners: push beyond generic insights. Tie advice to intent and cash flow timing, not just categories. Gate sensitive recommendations behind clear consent and explainability, and provide "show your work" views for regulated advice.
Instant payments are widespread, but pain points persist
Modern rails are changing behavior. E-wallets and account-to-account (A2A) now account for 58% of cashless transactions by value. That said, the experience isn't consistently instant where it matters most.
32% of consumers report slow processing, and 42% of business owners say slow payments hurt operations-rising to 50% in Malaysia and 52% in Vietnam. Refunds, insurance payouts, and cross-border transfers are the most underperforming use cases for speed.
Fix the gaps first:
- Automate refunds with instant rails and tight reconciliation to reduce support costs and churn.
- For insurers, use instant disbursements with claim-status triggers and fraud controls baked in.
- For SMEs, bundle A2A collections with real-time settlement and working-capital options.
Real-time cross-border moves to center stage
Project Nexus, led by the BIS Innovation Hub, is connecting domestic instant payment systems to enable faster, cheaper cross-border payments. Indonesia, Malaysia, the Philippines, Singapore, Thailand, and the Eurosystem are among those involved. For e-commerce and retailers, this expands addressable markets via QR and mobile wallet acceptance and should lift conversion. For remitters, payout becomes more uniform and simpler to run.
Learn more about Project Nexus at BIS
Stablecoins are gaining traction in regulated settings
Digital currency use is no longer fringe. 12% of respondents report weekly use. In Singapore, the SGD-pegged stablecoin XSGD surpassed US$8B in transaction volume by mid-2025. Clear rules are accelerating adoption: Singapore introduced a stablecoin framework in 2023 and is moving to formalize it through legislation, drawing in global players building enterprise use cases.
For finance teams, stablecoins can reduce settlement friction, enable programmable payments, and improve cross-border speed with clear controls. Prioritize policy guardrails, treasury handling, on/off-ramps, and merchant acceptance layers.
MAS stablecoin regulatory framework explainer
What finance leaders should execute over the next 12 months
- Embed credit at the point of need: BNPL and instant loans with income-aware risk tiers and premium bundles.
- Productize AI personalization: budgeting intelligence, spending insights, actionable nudges tied to cash flow events.
- Stand up an agentic AI sandbox: scoped APIs, consent frameworks, and policy-based transaction limits.
- Close "instant" gaps: automate refunds and claims with instant rails and end-to-end reconciliation.
- Prepare for Nexus-era cross-border: enable QR and wallet acceptance for inbound shoppers; streamline remittance payout.
- Pilot stablecoin use cases: B2B settlement, treasury sweeps, and programmable disbursements with clear accounting treatment.
- Data governance first: consent registries, model risk management, and audit-ready logs across AI and payment flows.
- Measure what matters: LTV/CAC uplift from embedded finance, refund time-to-wallet, cross-border success rate, and NPS impact.
Skills and tooling
If your team is scoping agentic workflows or AI-driven personalization, start with practical tooling and playbooks built for finance use cases. A curated list of AI tools for finance can speed evaluation and vendor shortlisting.
The takeaway is simple: embedded finance is mainstream, agentic AI is coming fast, and programmable money is the glue. Move early where your customers feel the friction-refunds, payouts, and cross-border-and build controlled AI and payment capabilities that compound over time.
Your membership also unlocks:
