Banks Urged To Move Beyond AI Pilots

Kuala Lumpur, July 8: Malaysian financial institutions must stop treating artificial intelligence as a series of isolated pilot projects and instead integrate it into their core business strategy if they are to unlock its full potential, industry leaders said at the Bank Audit Conference (BAC).
While banks have made steady progress adopting AI across areas such as customer service, compliance and fraud detection, experts said the next challenge is scaling those initiatives across the organisation without compromising governance, cybersecurity or customer trust.
The issue was discussed during a roundtable on AI readiness featuring Ecosystm Principal Advisor, BFSI Practice Sashikala Suresh, AICB Chief Executive Officer Edward Ling, ADO AI co-founder and chief executive Dr Ayesha Khanna, AmBank Group Chief Information Security Officer (Group CISO) Malini Kanesamoorthy, and Luno Asia Pacific General Manager Aaron Tang.

AI adoption gathers pace

Suresh said Malaysian banks were generally keeping pace with regional and global peers, but many organisations were still developing AI capabilities rather than embedding them into enterprise-wide transformation.
She said research conducted by Ecosystm involving 90 respondents from commercial banks, digital banks, Islamic banks and development financial institutions found that 44% of financial institutions were in the “developing” stage of AI readiness, while only 2% had reached an advanced level of maturity.
Rather than focusing on the small number of advanced organisations, Suresh said banks should view the results as evidence of a healthy pipeline of organisations progressing along the AI maturity curve.
“The way to look at this data is that you actually have a pipeline of readiness. Don’t look at just the 2%.”
She noted that many organisations were already building customised AI solutions but lacked documented business strategies to support them, resulting in fragmented AI activity instead of long-term organisational capability.
According to the survey, 68% of respondents reported measurable gains from AI investments, yet only 32% used financial metrics to evaluate those returns.
“AI is very different from traditional technologies,” she said, explaining that returns often emerge over a longer period and should be measured beyond immediate productivity improvements.
Suresh said some markets had already become more cautious about AI adoption, citing Australia as an example where organisations had stepped back to reassess their AI strategies and governance frameworks.
She said Malaysia had the opportunity to learn from those experiences by adopting AI with clear business objectives, robust governance and organisational readiness from the outset.

Moving beyond pilot projects

Dr Ayesha Khanna said many organisations remained trapped in the proof-of-concept phase because they attempted to automate individual tasks rather than redesign entire business processes.
“The first thing is that one needs to move away from this POC pilot phase. Nobody takes it seriously.”
Instead of approving every AI idea, organisations should identify a small number of strategic business priorities and select AI projects capable of delivering measurable improvements in profitability, growth and market expansion.
She also argued that enterprise transformation depended less on deploying AI tools than on redesigning workflows and involving employees throughout the process.
“Personal productivity increase is not equal to enterprise productivity increase.”
Rather than simply giving employees AI assistants, organisations should ask staff what they would do if repetitive work could be automated, allowing AI engineers and business teams to redesign workflows together.

Protecting data and building trust

Cybersecurity and governance also emerged as recurring themes throughout the discussion.
Kanesamoorthy said banks must evolve existing governance frameworks instead of creating entirely new ones specifically for AI.
“There needs to be a mindset shift,” she said, adding that AI should not be viewed as simply another technology upgrade.
She stressed that protecting customer data remained a top priority but should not become a barrier to innovation. Instead, banks should use techniques such as data masking, tokenisation and secure sandbox environments to allow AI experimentation while maintaining privacy and regulatory compliance.
The discussion also turned to the use of AI in handling sensitive customer data, with one audience member asking whether information such as identity card (IC) numbers and credit card details could be uploaded into AI tools.
Kanesamoorthy said confidential information should first be masked or replaced with dummy data before being processed.
“I said, look, you know that the company name is actually confidential, so why don’t you mask it or put dummy data? End of the day, your outcome is you want to actually do a comparison on the financial data, and do some analytics, so you’re actually not breaching anything.”
She said the same principle applied to sensitive personal information such as IC numbers and credit card details.
“AI can complement what we are doing today.”
Kanesamoorthy also rejected fears that AI would automatically replace workers, arguing that organisations needed to communicate clearly that the technology was intended to improve productivity rather than eliminate jobs.
“I think the messaging needs to be clear. AI is not here to replace you. It’s to support the agenda to achieve operational excellence.”

People remain central

The discussion also highlighted workforce readiness as one of the biggest barriers to AI adoption.
Ling said the future of banking would depend not only on technology but on people capable of using AI responsibly.
“The future of banking will not be delivered by technology alone. It will be guided by people who know how to use AI responsibly.”
He cited survey findings showing 79% of financial institutions faced shortages in AI-related technical skills, arguing that continuous learning should become part of every organisation’s business strategy rather than being treated solely as an HR initiative.
Suresh agreed that organisations needed to invest more heavily in change management than they had during previous waves of digital transformation.
She said AI adoption should not be left to technology teams alone, with human resources departments playing an important role in helping employees adapt to changing roles, identify skills gaps and build confidence in using AI across organisations.

Working with regulators

Tang said innovation in regulated industries depended on trust rather than speed alone.
Drawing on Luno’s experience in the digital assets industry, he said companies should engage regulators early rather than waiting for rules to be finalised.
“The companies that survived and thrived were actually those companies that were doing it in a trustful and respectful manner.”
He added that industry collaboration could help accelerate responsible AI adoption by giving regulators greater confidence in emerging technologies.
Suresh also warned against waiting for regulators to provide complete certainty before adopting AI.
“If organisations are actually waiting for regulatory clarity to start their AI journey, they’ve already missed the ship.”
Instead, she urged banks to work more closely with regulators, noting that 56% of respondents said they were not engaging enough with supervisory authorities.
“Regulation, by definition, is retroactive,” she said. “What you need to do is engage with industry bodies and regulators so that you can actually give them the real view of the world.”