Philippines banks can quickly catch upin AI race - Citi

Upgrade to High-Speed Internet for only ₱1499/month!

Enjoy up to 100 Mbps fiber broadband, perfect for browsing, streaming, and gaming.

Visit Suniway.ph to learn

“While maybe right now, institutions within the Philippines are not yet as advanced as some of the firms we talked to, I don’t think the gap is that big for them to catch up.”

MANILA, Philippines — Philippine banks and financial institutions may still be taking cautious steps toward artificial intelligence (AI), but the gap with more advanced markets can narrow quickly as firms move from experimentation to practical use cases, according to global banking giant Citi.

Helen Krause, head of AI and Data Intelligence at Citi Institute, said discussions at a forum for clients hosted by Citi Investor Services in Manila suggest that many local firms are still in the exploration stage of AI adoption, compared with some institutions in developed markets that have already moved to production.

Still, she said the shift could happen “really, really quickly,” especially as financial institutions gain a better understanding of what AI can do and as the technology becomes more capable.

“While maybe right now, institutions within the Philippines are not yet as advanced as some of the firms we talked to, I don’t think the gap is that big for them to catch up,” Krause said.

Her comments come as AI use in financial services shifts from back-office efficiency to more complex applications in investment research, risk management, reporting, trading workflows and wealth management.

From exploration to production

Globally, financial institutions are no longer asking whether they should adopt AI, but how to integrate the technology into their operations while keeping governance, accountability and data security intact.

The Bangko Sentral ng Pilipinas has already found signs of growing AI adoption among financial institutions. In its 2024 thematic review on AI and machine learning in the Philippine financial services industry, the BSP said 21 out of 48 respondents (44 percent) had deployed at least one AI system in production, while 29 respondents (60 percent) had included AI or machine learning in their institutional roadmaps.

The BSP said major use cases include fraud management systems, electronic know-your-customer and liveness checks, recommender systems, assistive AI or GenAI and credit risk scoring.

However, its review also showed uneven maturity across the industry, with data, models and tools scoring higher than governance, consumer protection and ethics.

For Krause, the uneven pace of adoption is not unique to the Philippines. Across markets, firms are on different AI journeys depending on size, priorities, technology budgets, risk appetite and management support.

“I think firms are on a different trajectory in terms of their AI adoption journey,” she said. “Some take a more cautious stance in terms of how they implement that.”

According to Citi Institute’s report “AI in Investment Management: Beyond Efficiency Gains,” AI adoption in the global investment industry has evolved from automating middle- and back-office functions, such as document processing, form filling, report generation and compliance monitoring, to front-office applications involving research, analytics and investment decision support.

The shift has been accelerated by generative AI and agentic AI, or systems that can plan, execute and adapt multi-step tasks toward a defined objective. Unlike a chatbot that responds to a prompt, an AI agent can be instructed to monitor a market development, check its impact on a portfolio and suggest possible actions for review.

Krause said this is where more advanced firms are beginning to move.

She gave the example of an agent set up to monitor market developments overnight. If an inflation shock occurs in the United States, the agent can summarize the event, assess what it means for a trading book or portfolio and flag changes in risk exposure.

“Another agent can look at portfolio construction and flag whether you are overly exposed to an inflation shock, identify positions that may need to be trimmed and suggest trades to reduce that exposure,” Krause said.

But she stressed that AI does not remove human accountability, especially in financial services. “We are not talking about a process that runs from the initial trigger all the way to execution. In financial services, someone still needs to be responsible. The firm ultimately bears the accountability.”

That accountability becomes more important as AI use expands into riskier areas such as investment recommendations, trading, portfolio construction and regulatory compliance.

The BSP has taken a similar view in its AI review, saying accountability for decision-making processes lies not with AI systems but with humans. It also said financial institutions need to address risks related to accuracy, hallucination, data quality, ethics, data privacy, cybersecurity and bias.

Krause said the most immediate opportunities for Philippine financial firms are likely to be internal and operational. These include document summarization, meeting transcription, report generation, data extraction, compliance monitoring and workflow automation.

Data privacy is even more critical in wealth management and family offices, where banks serve wealthy families and business owners with highly sensitive personal and financial information.

“Data privacy is absolutely non-negotiable. It has to be secure and respected with the utmost care,” Krause said.

Citi Institute’s family office report showed that only 22 percent of family offices currently use AI for operational tasks or investment analysis, although this was already up from 13 percent in 2024.

The report also found that 57 percent of family offices cited lack of internal expertise as the biggest barrier to technology adoption, followed by lack of awareness of available options at 34 percent and cybersecurity or privacy concerns at 28 percent.

Privacy, practical uses

Krause said some family offices still prefer Excel over AI if they are not confident that sensitive information will remain secure. Others may choose on-premise solutions, open-source models, trusted banking partners or a hybrid approach where some parts are built internally and others are bought from external providers.

The same “buy versus build” question also applies to banks and investment firms. Krause said the decision often depends on cost, benefit, speed to market, use case and data sensitivity.

“If a firm wants to ensure that its first-party data does not leak out, it may prefer to use an open-source model, bring it on premise and fine-tune the model internally,” she said.

For less sensitive and more general tasks, firms may use large foundational models. But for high-volume repetitive processes, smaller and cheaper models may be enough.

AI as everyday banking tool

Looking ahead, Krause said AI adoption in the Philippine financial sector could evolve quickly over the next few years, but only if firms choose the right use cases and develop the talent needed to use the tools responsibly.

“Three to five years is a long time in the AI era,” she said.

She said firms should keep pace with technology developments, but should not adopt AI simply because others are doing so.

“We see AI augmenting the work we do — helping automate routine tasks and improve productivity. While it’s still early to fully assess the impact of AI, we expect new job opportunities being created, with potential for those who embrace using it,” she said.

“Talent adaptation and upskilling is crucial. New employees need to be tech-savvy. The rise of AI requires a new approach to talent development. Firms must upskill their employees by providing relevant AI training because a human in the loop remains critical.”

For Philippine banks, the AI race may not be about who deploys the flashiest tool first. It may be about who can turn AI into a reliable part of everyday banking while keeping customers, regulators and employees confident that humans remain firmly in charge.