McKinsey hosted its first Asia AI & Data Summit in Bangkok early in 2025 against the backdrop of countries in the region embracing the AI and tech revolution. The event brought together over 100 leaders from a wide range of fields, including banking and financial services, to gain a better understanding of the growing AI markets in the Asia–Pacific region. The important role that data and tech play was a focus area in which delegates got a glimpse of the next potential AI frontiers.
In an interview at the summit, McKinsey experts Anke Raufuss and Renny Thomas discussed the transformative effects of AI in the financial and banking sectors and what an AI bank of the future could look like.
An edited version of their conversation follows.
What about the AI and technology revolution has been most transformative for financial services?
Renny Thomas: Banks have been using traditional and predictive AI for a while, but the advent of gen AI has opened up an entirely new paradigm: the ability to use aspects such as natural language and unstructured data. In my view, the biggest breakthrough has happened over the past 12 to 18 months, and that is multiagent systems. One of the limitations of gen AI has been its inability to do complex tasks. With multiagent systems, you can take complex workflows and automate them in a manner similar to how humans would do a task. Teams of machines working together can drive extreme automation and create very impressive customer experiences.
Anke Raufuss: I am excited about the innovation that AI brings, especially for customers who can now receive much better service. Their banks are basically in their pockets as an app, which also means that banks know what their customers are doing and how to serve them better. As for the banks, they now have more insight and transparency into what’s going on, which helps them better manage safety and risk.
What makes an AI bank of the future?
Renny Thomas: My view of an AI bank of the future consists of multiple parts. One aspect relates to full-stack AI to serve the customer more effectively and efficiently. I envision the bank of the future to be integrated, multimodal, and omnichannel so that the customer experience is seamless.
It needs to be at level-five personalization, often referred to as hyperpersonalization. This will enable traditionally siloed products to be bundled per the customer’s need. For example, if you have an investment product and want to borrow against it, it should be a seamless process. Or if you have a mortgage and want to offset that, it should be easy. Real product innovation and product bundling are important features, which should be able to be done with a much higher level of efficiency through extreme automation.
For me, all these would be feasible with the AI banks of the future, as long as they implement a comprehensive capability stack that contains four key capability layers: engagement, decisioning, core technology, and operating model. These four layers need to operate in sync and work in a consistent manner. It is not about small pockets of innovation and small pockets of AI wheels spinning in isolation, but all of these being connected in a manner that can drive the entire institution.
Anke Raufuss: Eventually, a bank’s business model is built on their customers’ trust. I think that with the new infrastructure and new opportunities, banks are now building trust in a digital way. Banks can gain that trust from customers by keeping banking systems safe and sound. Therefore, banks have technology risk management to consider, as well as how to use technology to keep their customers and employees safe. AI and technology provide banks with much better opportunities to understand who their customers are—not just what they want, but also whether they might be a danger or risk to the bank in the Know Your Customer (KYC) sense. Banks need to think about how to help customers avoid fraud, how to help them be safer, and provide better services.
What help from AI would you appreciate most from your financial services provider?
Renny Thomas: It would be the ability to be seamless across channels. For example, I should not have to contact a call center separately. If I’m talking to a relationship manager, he or she should know what my last interaction with the bank was and what my context is. If I show up at a branch, they should know who I am.
I would also like to have products that serve my needs as opposed to products that are more banking specific. For example, I would like a more seamless ability to save money and decide where to put that money, as opposed to merely buying a mutual fund or putting it in a savings account—not knowing why the bank made that call in the first place. Similarly, when I want to borrow money, I should be able to decide what I want to use that money for. It is about breaking through some of the jargon around these products. In other words, I would like the bank to serve my needs for saving money, making payments, or borrowing money, and I should be able to have flexibility about it.
Another aspect I would like is to be able to learn from other customers’ experiences, as opposed to having to rely only on the bank’s advice. I would like to connect with other customers so that when I plan to buy a product, I can get their opinions. It is not just about obtaining information about the service but also about the business of banking: Where do they save their money and what do they borrow? Basically, I would like more engagement.
Last, I would like to connect with my contacts and my money in a much easier way. I would like seamless service and an app that is not too complicated. I should be able to swipe and get something done, speak to the app and get it done, or text and get it done, as opposed to clicking on lots of menus and searching for something that I never seem able to find.
Anke Raufuss: I agree with the requirement of seamlessness of services. This might be because I’ve been inconvenienced a bit too often over the last few weeks when my credit cards have been canceled, just because I appeared to have had erratic buying behavior. I hope that the AI handling my transactions would have learned by now how erratically I typically operate and allow the transactions to go through. Hopefully it will adapt in the future.
