MANILA, 13 August 2026 — Artificial intelligence (AI) is moving from experimentation into financial-services workflows, shifting the challenge for banks from adoption to execution: where AI creates measurable value, how much autonomy it should have, who remains accountable for its decisions and how the operating model needs to change.

That was a central thread of The Asian Banker Finance Philippines 2026, where banking, regulatory and technology leaders examined AI in the context of wider changes to organisations, decision-making and customer relationships.

Opening the conference, Foo Boon Ping, president and managing editor of TAB Global, parent company of The Asian Banker, framed the transformation around three forces—connectivity, intelligence and trust. As AI increases the speed, scale and autonomy of financial decision-making, he argued that the challenge is not simply how quickly institutions can transform, but whether they can do so without weakening the trust and accountability on which banking depends.

AI has to move beyond the lab

Ana Aboitiz Delgado, president of the Bankers Association of the Philippines (BAP) and president and chief executive officer of UnionBank of the Philippines, compared AI with earlier technology shifts such as internet banking and social media.

She argued that AI is following a familiar adoption trajectory, but much faster. “Adoption of AI technology was not the hardest part,” she said.

Delgado identified three dimensions institutions need to get right beyond the technology stack: organisation, culture and ways of working. These encompass governance and decision rights; mindset, talent and leadership; and the workflows through which humans and AI collaborate.

She argued that banks need to move beyond the “lab model”, where AI remains within centres of excellence or isolated data-science teams. In an embedded model, AI operates across departments, model-risk specialists sit within risk functions, businesses and operations own automation pipelines, employees use AI-generated insights in their work, and applications have named owners and documented failure modes.

Delgado also differentiated assisted intelligence, where AI provides insights or recommendations but a person retains ownership of the outcome, from automated intelligence, where processes operate end to end with humans primarily reviewing or investigating exceptions. Her examples ranged from credit decisions and anti-money laundering alerts to fraud detection and real-time monitoring.

The test, she argued, is whether AI produces “actual impact”—measurable outcomes rather than simply more sophisticated technology.

She also stressed that AI should deepen rather than replace customer understanding. Models can process patterns and automate decisions at a scale humans cannot, but institutions still need to understand what customers are trying to achieve and why they behave as they do. Accountability, she argued, remains with the institution and its leadership rather than transferring to a technology provider, model or technology team.

BSP connects AI with money and trust

Florabelle M. Santos-Madrid, managing director of the Policy and Specialized Supervision Sub-Sector, Financial Supervision Sector at the Bangko Sentral ng Pilipinas (BSP), widened the discussion beyond AI itself.

Her keynote, “AI, Money, and Trust”, asked whether a more intelligent and innovative financial system would also become more trusted.

She placed that discussion against the condition of the Philippine banking system. As of June 2026, banking-system assets stood at PHP 31.1 trillion (about $509 billion), up 10.3%, while loans reached PHP 17.8 trillion (about $291 billion), up 11.9%. The non-performing loan ratio was 3.29%, with a 92.52% coverage ratio.

Santos-Madrid identified modern payment infrastructure, interoperable systems, digital identities, quality data and sound governance as foundations of a trusted financial ecosystem.

The BSP's approach, she said, is not to regulate technology itself but its risks while enabling responsible innovation. Its AI governance approach is organised around the STARS principles—sustainability, transparency, accountability, responsibility and security. Santos-Madrid described the framework as principles-based, proportionate and risk-focused rather than prescribing a uniform model across institutions.

She emphasised that greater technological capability does not reduce institutional responsibility. As AI becomes more deeply embedded in financial services, institutions need to address governance, human oversight, data quality, cybersecurity, third-party dependencies and accountability.

Santos-Madrid also highlighted collaboration across AI governance, sustainable finance, climate-risk assessment and cross-border payments. She cited Project Nexus, which aims to connect domestic instant-payment systems to enable faster, safer and more efficient international payments. Her remarks placed trust alongside technological capability as a central consideration in how financial institutions deploy new technologies.

SMBC moves towards end-to-end transformation

Mayoran Rajendra, managing director and general manager of the AI Transformation Department at Sumitomo Mitsui Banking Corporation (SMBC) Group, brought the perspective of implementing AI across a large international banking group.

Rajendra described SMBC's strategy as comprising three ambitions: improve, by accelerating operational efficiency and business processes; enhance, by augmenting products and services; and advance, by transforming business models around the future shape of financial services.

He also identified three levels of AI value creation: individual efficiency, process automation and end-to-end transformation. The first improves everyday productivity; the second introduces reusable AI components into workflows; and the third requires customer journeys and processes themselves to be redesigned around AI.

Rajendra said the third stage is the most difficult because banks cannot simply place AI on top of fragmented processes. Institutions need to start with business design and work backwards towards the technology.

He said SMBC had established a cross-functional structure involving 18 departments while investing in infrastructure, data and governance. Much of the work involved in becoming AI-ready, he said, was directed towards making data available and usable rather than simply acquiring models.

Rajendra also emphasised workforce enablement through everyday AI tools, training programmes, employee experimentation and a central AI hub. At enterprise scale, he highlighted resilience, business continuity, dependence on external frontier models and the cost of deploying AI across high-volume or mission-critical processes.

The experience illustrated the organisational and infrastructure work behind enterprise AI deployment. Rajendra identified cross-functional collaboration, governance and platform foundations, and safe and responsible scaling as areas SMBC is addressing.

Measurable value becomes the test

The subsequent Leadership Dialogue on “The new operating economics of AI banking” brought together Melchor T. “Mhel” Plabasan, senior director of the Technology Risk and Innovation Supervision Department at the BSP; Adrienne Heinrich, head of data science and AI at Maya; Rajendra; and Mark Philip “Mack” Comandante, founder and chief executive officer of Exoasia Innovation Hub and executive director of the Global AI Council Philippines.

The discussion examined four areas: measurable business value, AI autonomy, governance and accountability, and the operating model required for wider deployment.

Heinrich argued that institutions should measure AI through the underlying business outcome rather than create separate measures simply because AI is involved. In lending, that can mean conversion or delinquency; in customer support, resolution or satisfaction. AI, she said, is part of Maya's operating model, with use cases developed alongside the business owners responsible for the relevant performance measures.

She also illustrated how autonomy varies according to risk. Lower-risk customer engagement can tolerate greater automation than fraud or credit decisions with more serious consequences. Heinrich cited one AI-driven engagement application that she said generated about twice as many logins and more than 20% higher average revenue per user.

Plabasan said autonomy should be assessed according to the materiality and risk of an activity, data quality, meaningful human oversight and clarity of accountability. Higher-risk applications require stronger governance, while institutional responsibility remains even when AI supports or executes a decision.

He said the BSP is also applying AI internally, including work to automate aspects of supervision and augment rather than replace examiners, giving supervisors experience with some of the same implementation questions facing regulated institutions.

Comandante emphasised identifiable human accountability as AI becomes more deeply embedded in organisations. He also suggested that one test of whether an AI capability is strategic rather than transitory is whether it can survive a change in the underlying model.

Heinrich returned the discussion to organisational readiness. Technology may be the easier part of transformation, she said; processes, people and culture are harder. Maya has established AI champions outside the technology organisation to identify use cases, work with AI specialists and build capability across individual functions.

The opening session suggested that the competitive question around AI is moving beyond access to the technology. As models become more capable and widely available, what may prove harder to replicate are the capabilities surrounding them: proprietary and well-governed data, redesigned workflows, reusable infrastructure, employees able to work effectively with AI and clear decision rights.

The issue is therefore no longer simply how many models, agents or use cases a bank can deploy, but whether they improve customer, risk or operating outcomes; how much decision-making can safely be delegated; and who remains accountable when it is.

AI may increase the speed and scale at which financial institutions make decisions. It does not remove the need for judgement.

For banks, the more consequential question is not how much AI they deploy, but what actually changes—in their economics, customer outcomes and operating models—when they do.