AI in Philippine Banking: From Incremental Gains to Bold Bets
A group of banking leaders from the Philippines gathered in a roundtable hosted by Fintech News and OneConnect Financial Technology to tackle a simple question: have we been thinking too small about AI?
The consensus: many banks are chasing safe wins while missing the bigger moves that change cost structures, risk accuracy, and customer growth. If you lead a P&L or a function, this is your signal to set a sharper agenda.
The question on the table: Are we stuck with small wins?
Incremental gains feel safe-pilot a chatbot, automate a report, run a small model. But the upside is in rethinking end-to-end workflows where AI impacts risk, revenue, and experience at the same time.
The leaders in the room pushed for bolder targets, shorter build cycles, and governance that scales with adoption-not governance that blocks it.
Where AI creates measurable value now
- Credit risk: alternative data scoring for thin-file and informal income segments, faster approvals for SMEs, and early warning triggers.
- Fraud and AML: real-time anomaly detection, entity linking, and smarter alerts that cut false positives.
- Customer service: AI assistants for call centers and chat that deflect repetitive volumes while escalating the right cases to humans.
- Collections: dynamic strategies by customer micro-segment with prediction-driven outreach and settlement offers.
- Personalization: next-best-offer across deposits, cards, and loans with channel-level optimization.
- Ops automation: document extraction (KYC, onboarding, trade), reconciliations, and exception handling.
- Treasury and risk: scenario testing for liquidity and interest rate moves, faster stress-test iteration.
What separates leaders from laggards
- Problem-first approach: start with a P&L-linked use case, not a tool.
- Shipping rhythm: MVP in 6-10 weeks, iterate with real data, scale on evidence.
- Data foundations: clear owners, clean join keys, and approved feature stores.
- Model governance that enables: lightweight approvals for low-risk use, stronger gates for credit and AML.
- Talent model: small central AI team, embedded data product owners in lines of business.
- Vendor strategy: buy for commodity capabilities, build for differentiators.
- Security and privacy baked in: PII minimization, audit trails, and reproducibility.
Your 90-day plan
- Pick two use cases with clear ROI (e.g., collections uplift, fraud false positive reduction). Set a baseline this week.
- Stand up a cross-functional squad: product owner, data scientist, engineer, risk partner, business sponsor.
- Define success metrics and guardrails (e.g., approval lift with constant risk, answer accuracy, SLA adherence).
- Secure the data pipeline: access, quality checks, lineage. No data, no model.
- Run an MVP in production on a small segment. Prove value before scaling.
- Document model risk and monitoring-drift checks, bias checks, retrain triggers.
Risk, controls, and trust
Strong outcomes rely on clear model accountability, versioning, and ongoing monitoring. Keep humans in the loop where outcomes affect credit, fraud, or compliance.
For a solid reference on AI risk controls that regulators recognize, see the NIST AI Risk Management Framework here.
Metrics that matter
- Credit: approval lift at flat loss rate, time-to-yes, early delinquency rate change.
- Fraud/AML: detection rate, false positives, case handling time.
- Service: first-contact resolution, average handle time, CSAT/effort score.
- Ops: cycle time reduction, error rate, cost per case.
- Adoption: percent of decisions augmented by AI with monitored outcomes.
People and operating model
- Create AI product owner roles inside business units-accountable for value, not models.
- Set a model review board that meets weekly, not quarterly. Fast feedback is cheaper.
- Upskill managers to read model dashboards and act on them, just like financial reports.
Roundtable speakers
- Matthew Chen, CEO, OneConnect Financial Technology
- Jerry Ngo, CEO, East West Banking Corporation
- Manish Bhai, CEO, UNO Bank
- Lito Villanueva, EVP and Chief Innovation and Inclusion Officer, RCBC
- Gigi Puno, CTO, GoTyme Bank
- Nishy De Silva, Senior Vice President and Shared Services CTO, Security Bank
- Mike Singh, President, Tendo by Tonik
- Gus Poston, Co-founder, Netbank
- Mila Bedrenets, Chief Growth Hacker, Tonik
- Jonathan Uy, Head Of Strategy, Philippine National Bank
Moderated by: Vincent Fong, Chief Editor, Fintech News Network
Keep learning
If you're building your AI roadmap for risk, service, and growth, explore practical tools and courses built for operators. A good starting point is this curated overview of AI tools for finance: Complete AI Training - Finance Tools.
The signal is clear: small pilots are fine, but they won't move your cost-to-income ratio or risk outcomes by themselves. Pick the few use cases that matter, get them into production, and let results guide your next bets.
Your membership also unlocks:
