With banks already launching pilots to see where GenAI could strengthen operations, the technology promises to completely transform the way banks do business. FStech news editor Alexandra Leonards reports.

Behind the scenes banks have been actively investigating how they can take advantage of the latest developments in AI, with the technology – particularly generative AI (GenAI) large language models (LLMs) like OpenAI’s ChatGPT and Google’s Bard – gaining significant exposure over the past 12 months.

While variances between tech budgets or the size of data science teams mean that some are more advanced than others, many banks have reached a stage where they are able to identify GenAI use cases they want to explore further, with some even beginning to roll out pilots or prototypes to identify where the technology best fits into their operations.

Some have already gone further, with Capital One and JPMorgan Chase already leveraging GenAI to augment their AI-powered fraud and suspicious activity detection systems. This has already delivered positive results, with an International Monetary Fund (IMF) whitepaper praising it for “a significant reduction in false positives, a better detection rate, reduced costs, and improved customer satisfaction”.

Clearly then GenAI is poised to play a significant role for banks of the future, with its arguably limitless nature signalling a vast array of new possibilities that are already beginning to reinvent the banking industry.

Finding the right use cases

NatWest Group is one UK-based organisation currently exploring the technology, having already assessed around 140 different use cases. The bank has whittled its idea pool down, grouping together those that could work under the same infrastructure.

But choosing the right use case isn’t always easy, with some organisations struggling to identify where they can make the most out of the technology.

NatWest head of digital journeys Chris Waring, who works on the business banking side of the industry, tells FStech that banks need a lot of discipline when it comes to finding the right GenAI-powered tools and solutions.

“We've been as creative and as broad as we could be at this stage to look at different ideas and solicit them across the different businesses,” says Waring, explaining that the technology is being experimented with across everywhere from retail and investment banking to servicing operations and the digital department in which he works.

Jeff Tijssen, global head of FinTech at international consultancy Bain & Co., says that while it is relatively easy to come up with 250 different potential use cases, the difficulty lies in identifying a framework for prioritising where to start.

“It's incredibly important to effectively build those foundational capabilities and start small before you start to explore the bigger, more complex use cases,” he explains.

Waring explains NatWest has taken a classic strategic approach which involves scoring each idea based on how much value it would bring to customers or the business. Once this base score has been evaluated, the bank identifies any risk involved and how to mitigate it, and then attempts to figure out how strong the guardrails around the technology would need to be to ensure that it is used effectively.

The organisation also considers ‘reusability’ when deciding which cases are worth pursuing – essentially looking at whether a GenAI tool built for one function or area could be used in another.

With all other aspects factored in, the bank decides how technically feasible an idea would be to bring to market in a short space of time.

“For the time being, some of the ideas are more feasible than others,” continues Waring. “You wouldn’t just direct customers onto a carbon copy of a commercial LLM such as ChatGPT and use it live, but you would put some controls around that and ensure you build tooling on top to ensure the customer gets the right customer experience.”

Where is the technology already being developed?

NatWest is currently looking at how it can incorporate GenAI into its customer-facing chatbot Cora, which it has been developing over some time, as well as across corporate and industrial (C&I).

At the time of writing, the bank is assessing the ‘commonality’ necessary for the infrastructure behind the technology to support the different use cases it has identified. In the meantime, it is experimenting with pilots, such as a tool being used in the background of Cora to generate ChatGPT-style prompts for staff directly talking to customers.

“I know at the very top end of the book our investment banking colleagues and corporate are using [GenAI] to take information out of company accounts and combine that with trends in the market to be able take over some of the laborious work that an analyst would often do,” says Waring, adding that the bank is prioritising rolling out certain measures to ensure that the information coming out isn’t blindly accepted.

Additionally, the bank has plans to launch a Cora-like chatbot for its internal operations.

“There’s a whole vast array of information that we have access to internally and we’re trying to centralise that and draw those insights out to assist in helping a customer with unusual edge cases,” explains Waring. “And I do think that when it comes to those edge cases it’s a bit of a passion of mine that we solve them well.”

He continues: “One of the big advantages of the big banks is that we can deal with those unusual edge cases where some of the smaller banks may focus most of their attention on that large group of people that have a very ‘to be expected’ journey. We should focus on serving customers with a delightful journey, irrespective of whether it conforms to the norm. Being able to use AI to help with that and route customers appropriately I think is a really interesting use case."

Deutsche Bank is also currently testing GenAI tools, and in late 2022 announced a high-profile multi-year innovation partnership with computing giant Nvidia to accelerate the use of AI and machine learning (ML) in financial services.

“Examples include software code development, AI chatbots for internal and external use cases and performing faster risk calculations,” says Christoph Rabenseifner, chief strategy officer, technology, data and innovation at Deutsche Bank. “Our innovation team has worked closely with all business divisions to assess the demand and business cases for AI.”

Is regulation holding banks back?

Any discussion around AI will inevitably lead to the same points around murky ethical and regulatory guidance. If you ask most bankers, explains Bain & Co’s Jeff Tijssen, they want to see a clear regulatory framework “yesterday, rather than today”.

He says that even the largest banks in the world have faced regulatory challenges, adding that they “don’t want to make the same mistakes again.”

“I think there’s also significant concern around reputational damage,” continues Tijssen, explaining that financial services firms are proactively asking for regulation that can provide them with a clear framework for AI.

As regulated entities, this is especially true for banks who are keen to understand what they can and can’t do with the technology.

“Therefore, there’s definitely some hesitation around what is and isn’t possible because banks don’t want to do something now and get their fingers burned, or the following day get a call from the regulator that says, ‘you’re not allowed to do that’,” adds Tijssen.

But NatWest’s Chris Waring says that a lack of regulation hasn’t stopped the bank from looking at use cases or experimenting internally. He says that while the way regulation may come to bear is not yet clear, in the absence of a strict framework, to prepare for the possible risks of the technology it’s important that banks hold to the principle of improving the experience for customers whilst making it safer for them to do business with them.

Ultimately, he explains, NatWest wants to use the technology to enhance experiences, boost productivity, and make sure it is giving customers what they want by way of its financial services.

“You want to do that so that you're only additive in terms of the experience and so my sense is that these things are needed to make sure that it's protected as an industry,” he explains. “I think we have a really strong sense of what we should be doing in AI in terms of trying to anchor ourselves to those principles. I think it's useful to have that across the industry.”

Hyperpersonalisation

A trend for banks that looks set to emerge in the not-too-distant future is the use of GenAI technology to roll out so-called ‘hyperpersonalisation’, which promises to present customers with a highly customised and personalised experience throughout their journey.

Deutsche Bank’s Christoph Rabenseifner says that the bank is already seeing improved personalisation through existing AI tools.

“For example, in wealth management we already use algorithms to monitor risks, make regular recommendations and learn from the reactions to them,” he explains. “We anticipate that generative AI and large language models will take this experience to the next level.”

Jeff Tijssen of Bain & Co says that AI will effectively allow a bank to provide a hyperpersonalised experience to every single customer, whether they are a pub owner, a large corporate or the consumer.

“It’s a shift from being very reactive to being proactive and predictive,” he explains.

Chris Waring from NatWest agrees that GenAI-powered hyperpersonalisation will not just impact individual consumers, but also businesses, from small companies to large enterprises.

Based on his experience working on the commercial side of the bank, Waring says that the most important consideration is assessing how a company interacts, adding that hypersonalisation overlaps with things like embedded finance.

“The traditional way of thinking about things is that you want a loan to be able to do X, Y and Z, but actually hyperpersonalisation for me focusses more on the ambition and the goal the customer is trying to achieve,” he says. “Hyperpersonalisation for me is not just about customising how you interact with us in a way that you want to, but it’s also about being able to think about how to engage on those goals.”

He continues: “Historically a manager is sat in front of you, and you have those kind of discussions periodically. This allows us to do it continuously, so you see how the cash flow is developing and you see how the business might have new opportunities.”

Ultimately, bringing together these different data sources could help a bank in structuring a finance package in the future.

“Essentially, customers will come to the bank not just because they want to buy something, but because they want to try and identify how that might work, how they might finance it, and what they need to make that happen,” says Waring. “You can have all those data impacts evaluated and structured in a way that it's a lending product for you, it’s structured in a way that makes sense for you and your business rather than shoving you into a pigeonhole of ‘this is a this is the only product we can offer you and it kind of matches you’.”

The Co-pilot

Waring also talks about the possibility of using the technology to carry out credit modelling for larger businesses. When a company with a turnover surpassing £100 million wants to borrow large sums of money, for example, a specialist credit expert will assess the request.

“You can imagine that in order to really thoroughly investigate all the minutiae in someone’s reports, plus all the industry trends and everything else, it’s near impossible,” he says. “So, I see the potential for a co-pilot type solution that pulls in all the data and analyses it.

“Essentially, banks will have a co-pilot next to the credit officer to assess that type of structure and possibly suggest alternatives to the way that you might arrange the financing to help the customer achieve their goals.”

Jeff Tijssen says that the technology provides a big opportunity for relationship management.

“If your bank focusses on small businesses or corporates, to be able to constantly understand what’s going on with that company through social media trends and beyond puts you as a bank in a position to be a lot more proactive as opposed to reactive, which is where most banks still are today,” he explains.

Waring thinks that the technology will move beyond just interactions between GenAI and a colleague or customer.

“I do think the next stage of that evolution might be GenAI to GenAI and how that might produce an interesting interaction,” he says. “We have Bank of APIs which is pretty market leading in what we give developers access to, so I also think it would be interesting to see how using AI to interrogate financial information could be a next step."

He concedes that “It’s not something yet on the radar,” but notes that “it's one of those things that clearly will start to develop so that an API is not just requesting a single piece of data in a structured and predictable way, but also looking at how you can then start to access that a little bit like a human would interrogate information in a set of accounts.”

The profit centre of the future

Tijssen says that despite billions of investment in digital transformation, a big part of financial services is still very manual. But he thinks generative AI will unlock more innovation, higher levels of productivity and higher levels of hyperpersonalisation.

“The ability for you to launch new products and services to rapidly respond to trends and developments in the market will give you a massive boost and really accelerate your time to market for doing those kinds of things,” he says.

He adds that the technology could give banks a unique competitive advantage, with those “really doubling down” potentially using it to create a whole new revenue stream for the organisation.

“And we're already seeing this, of course, with banking as a service and better finance where banks are really spending quite a bit of money and effort on building out new ventures that effectively turn a cost centre into a profit centre,” says Tijssen. “If you've invested in building a state-of-the-art AI platform for your own purposes, why not open it up to a whole new revenue stream for the organisation.”

Deutsche Bank’s Christoph Rabenseifner agrees that there are opportunities provided by GenAI that can span across the full breadth and depth of the bank’s activities.

“Over the next five years, AI, ML, and GenAI will transform every part of banking, from internal processes to interactions with our customers and vendors,” he concludes.

With banks already exploring the many opportunities that GenAI can offer the market, it’s clear that as the technology develops further it will have a huge impact on the sector and fundamentally reinvent the way it does business.

This however will all rely on the rules and guidance put in place by regulators here in the UK and across the world. It is of the utmost importance that all parties come together to create a framework for GenAI which ultimately allow banks to deliver their customers with the safe, secure, and delightful journey they deserve.

All eyes now turn to Bletchley Park in early November for the first ever global summit on artificial intelligence safety and the crucial decisions made by the world’s leaders therein.

Share Story: