Some tasks that once took hours could be completed in minutes, giving organizations more choice in balancing speed, accuracy and experience. AI also excels in monitoring and predicting regulatory change. For global banks, staying on top of shifting regulations is a huge challenge, and missing updates can jeopardize compliance and reputation. AI algorithms can analyse text for semantic differences, flagging potentially subtle but important changes.
For example, the FDIC recently clarified how banks can use data in the public domain as part of customer identification. AI could identify this nuanced change and flag its significance far quicker than a manual review in some situations.
By applying AI to regulatory monitoring, banks could detect subtle updates, interpret their impact and adapt processes more quickly, to ensure compliance. Corporate banking is also under pressure to modernize online experiences in keeping with retail banking and bring consumerization to the customer experience. Currently, there are too many manual processes that customers must navigate. AI can help streamline the experience by digitizing assets to serve higher customer expectations.
Good Data Is Paramount
AI is essentially the automation of prediction, using data to anticipate and model outcomes. Clean, structured and context-rich data is essential for accurate results, particularly in complex, multi-jurisdictional environments.
In KYC, AI improves efficiency by automating the prediction of what information is needed and where to find it. Once collected, the true value lies in correlation — building models that turn inputs into outcomes. To support this, organizations must ensure data is consistent across systems, auditable for governance and regulation and contextualized to reflect regional rules.
But risks persist. Models will never be 100% accurate, and hallucinations occur when outputs deviate from expected reality. The key is governance ensuring data is consistent across systems, auditable and contextualized to regional rules. By combining automation for data collection with AI for correlation and modelling, banks can unify fragmented datasets, improve accuracy and build the strong foundation required for trustworthy, scalable outcomes.
Human review remains essential to validate anomalies and ensure efficiency and trust stay in balance.
Agentic AI Connects Systems in A More Structured Way
Until now, most banks have used AI to complete specific tasks in disparate silos, often relying on additional API integrations to make the systems communicate. While there’s nothing inherently wrong with connecting systems this way, it can be time-consuming, costly and sometimes the workflows themselves need to be redesigned rather than simply linked.
Agentic AI represents a shift. It adds structure to AI models and embeds them into broader software ecosystems, enabling systems to communicate more efficiently. Unlike generative AI, agentic AI breaks down large, complex prompts into smaller sub-parts, adding structure that improves accuracy and precision. In practice, this often reflects the structure of existing or redesigned processes and still relies on APIs to connect systems.
The real advantage lies in automating the control of this infrastructure. By orchestrating workflows across systems, agentic AI delivers efficiencies that would otherwise require manual oversight and integration. Yet, human involvement remains critical. A human-in-the-loop approach ensures decisions are reviewed, anomalies are flagged, and accountability is maintained, recognising that while humans are fallible, their judgment provides the oversight needed to balance automation with trust.
Keeping Data Safe in the Age Of AI
AI has come under scrutiny for security gaps that can put client data at risk. While fraudsters and bad actors may try to steal sensitive bank data through AI’s security gaps, banks can fight fire with fire by using AI to keep data safer.
I’m on the board of the RegTech Association, which works with regulatory tech vendors, regulators, the SCC and the like to facilitate conversations about how people can respond to regulations as efficiently and accurately as possible. We have seen that ethical and responsible use of AI is incredibly important to banks and regulators. In talking to leaders over the last 18 months, I have seen attitudes shift from super conservative to a sense of “We have to do this.” Doing nothing is not an option.
From a governance and control perspective, banks often prefer to limit personal productivity tools to a small set of mainstream cloud providers, such as Google and Microsoft. In these cases, the AI operates within the bank’s firewall, reducing complexity and maintaining clearer lines of oversight and accountability. When it comes to using customer data, banks already have rigorous risk management for vendors and data security. The key to using AI safely is to take this risk management even further and adhere to emerging security standards like the ISO 2701 and ISO/IEC 42001, an international standard for using, maintaining and continuously improving AI management systems. This regulation is designed to ensure responsible development and use of AI systems that keep data safe.
ROI Is Emerging, but Still Evolving
While the full ROI from AI adoption is still unfolding, early signals across industries are encouraging. Organizations that have begun replacing manual, repetitive processes with automation and AI are already seeing tangible benefits, not just in cost savings, but in agility, customer satisfaction and strategic growth.
For example, ING CCO and Chief Transformation Officer Marnix van Stiphout recently shared in a McKinsey & Company interview that his organization is already seeing ROI in improving customer service, marketing, lessening the documentation burden in KYC processes and predicting customer churn that allows for more proactive and personalized outreach.
Banks Cannot Wait
However, ROI is not one-size-fits-all. It depends on the quality of data infrastructure, maturity of AI integration, clarity of business objectives and the ability to measure outcomes beyond cost savings.
The true ROI of AI will unfold in the future, but banks cannot wait to see where the numbers fall before taking action. They must act now on AI to keep up with competitors. Missing the boat on AI would be catastrophic to any bank’s future growth.
