Across financial services artificial intelligence (AI) is no longer the stuff of the future. It has been embraced by banks supported by third party fintechs, providers, and technology departments within banks keen to use their own resources to drive their use of the robotic process automation (RPA) and machine learning (ML).
And while some regions are more receptive than others, generally speaking AI has become part and parcel of many bank’s infrastructure – providing solutions from the front, middle and back offices, to becoming the basis for enterprise-wide software requirements. According to a report by consultancy Autonomous NEXT, 2.5 million financial services employees are currently exposed to AI technology. That looks set to increase.
“Organisations are a lot more receptive,” says Parth Desai, founder and CEO, Pelican, an organisation that applies artificial intelligence technology to payments and compliance with a number of the world’s largest banks. “I think the emerging countries are always open to trying out new technologies. At the same time the US is more open than Europe, generally regarding technological developments although Europe is catching up. People are willing to give it a shot because AI has been talked about over the past few years.”
But while AI has entered the common parlance among bank technology officers in recent years, it can still be a challenge for vendors when they knock on the door.
“There is still a challenge,” says Desai. “You need to have AI working in a real time environment which is very different from a prototype or proof of concept that you can come up with in one or two months. It needs to be integrated with systems it has to work constantly 24/7.”
That may seem like bad news for vendors, but the problems identified by banks this year suggest assistance is needed. According to a recent report by consultancy Turner Little, 89% percent of banks cite cyber and data security as a top priority this year, with 85% planning to implement a digital transformation program. As much as 77% of the sector said it must improve risk management processes. At the retail end, consumers have cited “unclear, complicated language and charges” as a frustration in traditional banking, and 67% of banks have said they aim to invest in technology to expand their ability to acquire, engage and retain customers.
And AI can have a direct impact on those problems. According to an Accenture survey, 71% of global banks believe AI can help build customer trust and confidence, 63% believe it will help optimise operations, and 62% are looking to AI to improve compliance procedures.
From biometrics to conversational interfaces, workforce and compliance automation, as well as product manufacturing in investments and insurance, the scope is huge. In the front office, integrating customer data and account actions with AI and applications that can hold conversations with clients – as well as support staff – are already in use. In the middle office, regulations have become much stricter in recent years, driving a trend towards real-time, AI oversight, risk management and know-your-customer (KYC) systems. In the back office, AI’s ability to assess with more depth different types of data sets and inform product manufacturing has been identified as revolutionary. In the same Autonomous NEXT report, the consultancy predicts that $490bn could be saved in the front office, $350bn in the middle, and $200bn in the coming years, across banking investment management and insurance. Banks are certainly aiming to get in on the act.
In 2016 Bank of America launched Erica – an intelligent virtual assistant, or robo advisor, that uses cognitive messaging and predictive analytics to offer financial guidance. The bank isn’t alone in driving the use of robo advisors, with institutions from UBS, Credit Suisse and JPMorgan just a handful of those aiming to apply AI to retail investing services.
JPMorgan has also announced its Contract Intelligence (COiN) platform, designed to “analyze legal documents and extract important data points and clauses,” using AI and machine learning technology. BNY Mellon is focusing on RPA to help improve its operations and reduce costs, while Wells Fargo established an Artificial Intelligence Solutions unit in February this year and Citibank is said to have a number of partnerships with leading edge AI providers.
Payment provisions
A number of fintech providers are focusing on the technology, and there’s some obvious benefits within the payment space and where the bank comes in direct contact with the customer that AI applications are being driven. Payment provider Finastra’s chatbot offering is integrated into the front end in order to serve retail clients. The technology can react to more than 200 most asked in-branch questions, according to the company. It’s also working with third party fintech organisations to develop facial or voice recognition tools to onboard customers without them having to go into branches, and voice-activated payments.
“When you’re sat on your sofa at home and you want to talk to Siri or Cortana or Alexa and you will be able to initiate payments,” says Martin Häring, Finastra’s chief marketing officer. Pelican’s Desai agrees that AI’s application can be transformative, helping banks reduce the cost of transaction banking payments by between 50-80%.
“Where there is human intervention, that’s where AI can be applied,” he says. “There is a lot of manual work because information is provided by the end user, normally in free format text. That needs to be converted into structured codes that the computers can understand.”
Pelican’s natural language processing software can be integrated holistically from the point of the end user backwards into the rest of the organisation’s worklflow, says Desai. “Once that’s done there’s other applications, such as automating customer onboarding, routing, flow, scheduling of the payments. That’s all streamlined and straight-through processing (STP). Then there’s a lot of reduction in the investigation and errors, queries, all that type of thing,” he says.
As banks have pivoted more towards services in the past twenty years, so too have various departments found applications for AI recently. One such instance is within treasury services. Bank of America Merrill Lynch (Baml) is applying artificial intelligence across its businesses, and notably within its Global Transaction Services (GTS).
“[I]t’s kicking off a customer service race with our clientswhere we are streamlining our operations,” says Dean Henry, head of innovation within GTS at Baml, who suggests the organisation is pursuing AI because of client expectations as well as the need to stay ahead of the competition. “Today, clients have higher expectations around faster turnaround times or predictive interpretation of issues happening or before they even get presented to them.”
Post financial crisis, banks and financial services providers have felt pressure to hunt out arbitrage while there’s been slimmer pickings across markets over the past few years. Changing market dynamics have heightened the need for better insights in the data those institutions have at hand, and AI is providing a route to that.
“Our clients are well aware of the value of data and they too are looking at AI,” says Henry. “They’re looking at how they take data and use it better within their own treasury operations.
“That goal is heightened right now because we’re finally entering a higher interest rate environment,” he adds. “The opportunity cost of not having a complete picture of a treasurer’s cash position is much steeper today than it was three or four years ago. So clients are asking for products that will help them improve their cash flow and understanding of their core business.
Third party vs in-house intelligence
With the Open Banking revolution and PSD2 driving banks to integrate more with fintech providers for fast-moving technologies such as AI and machine learning, it seems logical that more financial market participants would embrace the idea of third party arrangements over such specialist developments. Not everyone is convinced, however. Hans Tesselaar, executive director of the Banking Industry Architecture Network (Bian), the not-for-profit industry association, says banks want to move away from using fintechs in order to close off their systems.
“Over the last two or three years I’ve seen a desire to become less dependent on the vendors, but the question is how,” he says. “So a lot of banks are saying ‘OK, we have something in place, we need to move on. How do we do that?’”
But those in the market believe technologies such as AI are forging greater integration between fintechs and banks. For GTS’s Henry, the relationship between his business unit and fintechs is important.
“There are absolutely instances where we bring in solutions from the outside – but obviously not before testing it and making sure it’s safe and secure for us and our clients,” he says.
“We bring in outside technologies but then at the same time we are building and shaping our own. So on the retail side of our bank, Erica is a conversational AI tool that our US retail banks rolled down and we are all learning – across the entire bank – about conversational AI and how it can be used from that experiment. That technology was built in-house with a combination of different providers of whoever’s the best of the industry and the various aspects that were needed. Within GTS we’re trying to partner with the best – it’s important for our clients to know that.”
Finastra’s Häring echoes that sentiment, going a step further to suggest the banking industry could be a reflection of the technologies at hand as well.
“I would predict that over the next ten years a bank’s brand is going to become less important,” he says. “In a world where a bank is a platform, you’re really looking at bank as services through mobile or the front end, you might not realise which firms are serving things in the back end. So in the future to deliver world class banking services in the form of financial applications or in the form of better connectivity between financial services or large corporations I think that’s the world of tomorrow.”
Bian’s Tesselaar suggests the trend he has seen has made the fintech market more refined, offering more niche solutions. As fintech becomes more competitive, new technologies should as AI could be driven faster within unique product offerings. “Some vendors are breaking up their systems and offering components, so you can only purchase parts of it and there’s open integration on top of it,” says Tesselaar. “Some others are into open development environment for third parties who can add to their existing solutions.”
Changing workforce
AI’s integration into the banking network could be leading to a resource shift, and naturally leads to questions surrounding the workforce. The technology’s capabilities stretch to a framework in which humans operate alongside intelligent software, some jobs are more at risk than others. According to 79% of Accenture’s survey respondents, AI will work alongside humans within their organisations as “co-worker, collaborator, and trusted advisor” within the next two years. They also expect “the majority of bank-customer interactions will be conducted via AI” over the next few years.
“There are jobs that are more likely to fade away, or at least to get reduced,” says Finastra’s Häring. “The whole brokerage segment is an area where probably 90% of that job could be replaced by certain AI elements.”
But for Baml’s Henry, AI’s relevance may not have a huge impact on his workforce. “We don’t think it’s going to have a big shift in terms of number of people working in treasury services. The shift will be in the expectations for our customers around service,” he says.
What is clear is that the uptake of artificial intelligence could be driven by regulations, either directly, such as has been the case with Open Banking and PSD2, or indirectly, such as was the case with the second Markets in Financial Instruments Directive (Mifid II), which forced many financial institutions to rely on technology for weighty compliance requirements. But with the changing relationships between fintechs and banking institutions seemingly moving quickly, regulators have become sceptical around AI according to Häring. After all, AI relies squarely on data, and regulations such as GDPR are forcing firms to become stricter with how to treat that information.
“The regulators now are taking a look at how firms run their data centres and their data security,” says Häring. “This is in contrast to Open Banking and letting fintechs access that data and allowing fintechs turn that data into recommendations and better customer service. This is the ying yang that we currently see in the industry.