Could AI work at your community bank?

No longer the stuff of science fiction, artificial intelligence and machine learning technology are now in use at community banks for a growing number of internal and customer-facing applications.

By Karen Epper Hoffman • Illustrations by Mark Allen Miller


Taking a page out of the show Black Mirror, coverage of today’s technology paints a picture of a near-term future where robots powered by artificial intelligence overtake the human workplace and insert themselves into every element of daily life.

While machine learning and artificial intelligence, better known as AI, are indeed being applied in business technology, they aren’t necessarily a prelude to a dystopian future. Instead, businesses—community banks included—can use this technology to augment jobs rather than replace them. AI could even improve the customer experience.

Indeed, a PwC study shows that 52% of financial services industry executives are currently making “substantial” investments in AI, and 72% of business decision-makers believe AI will be the business advantage of the future. Matthew M. Speare, the executive vice president and chief information officer of $4 billion-asset Carter Bank & Trust in Martinsville, Va., says his community bank is leveraging this technology to create an enhanced customer experience and drive down costs on the back end. “With the robotic process automation [of AI], it can help a bank be more nimble and defensible,” Speare adds.

The aggregate potential cost savings for banks from AI-based applications is estimated at $447 billion by 2023, with the front and middle office accounting for $416 billion of that, according to Business Insider Intelligence. Eight out of 10 banks are highly aware of the potential benefits presented by AI, according to an OpenText survey of financial services professionals. In fact, three-quarters of banks with more than $100 billion in assets are currently implementing solutions enabled by AI, while nearly half, or 46%, of banks with less than $100 billion in assets are doing so, according to a UBS Evidence Lab report.

AI-based solutions have already gained a footing in banking with chatbots and payments fraud detection. Banks are also realizing they can use AI to create stronger connections with customers and third parties alike. Carter Bank & Trust has been using this technology in its unsecured consumer lending to support discovery and lead to “intelligent automation.”

“With machine learning on the back end, as a bank, we didn’t want to have 30 people doing data entry,” Speare says. “In the machine learning space, to massage data and pull from multiple sources and put into the systems we needed it to go into, that helped.”

“With machine learning on the back end, as a bank, we didn’t want to have 30 people doing data entry. In the machine learning space, to massage data and pull from multiple sources and put into systems we needed it to go into, that helped.”
—Matthew M. Speare, Carter Bank & Trust

Carter Bank & Trust hasn’t implemented AI technology in customer-facing applications, such as online chatbots, but it does rely on AI for its internal operations. Speare points out there are potential applications for supporting Bank Secrecy Act (BSA) and anti-money laundering (AML) efforts at a community bank. For example, the bank is using machine learning with its regular research to determine if a customer is participating in high-risk behaviors.

Marc Butterfield, senior vice president for innovation and disruption at $20 billion-asset First National Bank of Omaha in Omaha, Neb., says his community bank has been an early adopter of AI, particularly with consumer lending. “We recently partnered with a Silicon Valley fintech called Upstart to decrease loss rates and increase approval rates by using the platform’s AI technology,” Butterfield says.

As a sizable agricultural lender and credit card bank, First National Bank of Omaha has “a lot of information on customers to sort through, to offer them good solutions,” with which AI can help, according to Butterfield. As he sees it, AI helps community banking “get back to its roots … providing advice and guidance to customers.” For example, this technology can be used in compiling data to inform a customer about their everyday spending and potentially encourage proactive positive changes. “Because there are opportunities emerging to provide better customer experiences in financial services,” he adds.

How one community bank is putting AI to work

Marc Butterfield

First National Bank of Omaha’s “journey” into artificial intelligence began more than two years ago with the goal of better understanding how it can add value to the bank. Today, the $20 billion-asset community bank in Omaha, Neb., is experimenting with AI to make customer service improvements.

“Banks are using online chat to begin the conversation [with customers and prospects] before handing that person off to a live employee,” says Marc Butterfield, senior vice president for innovation and disruption.

On the back end, First National Bank of Omaha is considering AI for anti-money laundering (AML) and know your customer (KYC) compliance. It’s an evolving use case, Butterfield says, adding that while compliance can certainly use AI and machine learning technologies, compliance employees understandably have some reservations about AI and machine learning tech.

“Some risk averse-minded individuals have more of a mindset of ‘this is the way we’ve always done it,’” he adds. “That’s not a knock on them. It’s their job to protect the bank.”

Big investment, big payoff?

It would be easy to see AI as overhyped given all of the talk about its potential. But experts believe that potential is enormous. Implementing AI-based technologies could save banks as much as $1 trillion in revenue by 2030, according to research from Pymnts.com. And AI developments are expected to add $16 trillion to the global economy by 2030, according to data analysis from Emerj.

That potential has already led to investment from the financial sector. A 2017 PwC study found that 52% of bankers said they’re currently making “substantial investments” in AI, and 66% said they expect to be making substantial investments within three years. Research shows that few small and mid-sized financial institutions have made investments into AI-enabled technology. Earlier this year, Cornerstone Advisors surveyed senior executives at U.S.-based mid-size banks and credit unions of between $250 million and $50 billion in assets for its What’s Going on in Banking 2019 study. The financial industry consultancy found that a mere 2% of mid-size financial institutions have already deployed chatbots, machine learning or other AI technologies, and only 5% have implemented robotic process automation.

“Artificial intelligence use is rare when compared to machine learning,” says Quintin S. Sykes, managing director of Cornerstone Advisors of Scottsdale, Ariz. “I’m distinguishing between the two, as machine learning leverages historical data to train models for supervised learning, where AI attempts to gain insight into data on its own, or unsupervised learning.”

Sykes sees machine learning development in banking happening in several areas. These include fraud detection, where neural networks have been used to score credit and debit card transactions for fraud for decades, and fraud prevention, where machine learning models score the risk of logins and money movement activities in digital banking solutions. It’s also being used to underwrite consumer loans, where consumer credit risk models are being tuned using machine learning.

In marketing, it’s used to produce next-best product and attrition models and generate alerts to prompt customer outreach for cross-selling and retention. Machine learning is also used with optical character recognition (OCR), where the MICR (magnetic ink character recognition) name line and amount on checks and other handwritten documents are scanned is a standard practice.

It’s all about the data

But the current is speeding up. “Banking in the 21st century is no longer just about managing investments and money, but more about the way financial institutions work with their data,” says Yogesh Pandit, the founder and CEO of Hexanika, a U.S.-based company that automates business processes using AI and big data.

“Banking in the 21st century is no longer just about managing investments and money, but more about the way financial institutions work with their data.”
—Yogesh Pandit, Hexanika

“In lieu of this,” Pandit continues, “technologies like artificial intelligence, big data, blockchain, cloud, security, IoT [Internet of Things], robotics and robotic process automation have started playing a key role within the financial industry.” Pandit sees banks wanting AI and machine learning to help them sift through their stockpiles of data, if nothing else.

“Financial institutions in the U.S. and globally need AI particularly to store and properly manage the large amounts of data that they need to maintain to gain analytics, insights and reporting,” he says. And complying with the new regulatory standards like the BCBS 239 principles, which mandate financial institutions to provide data accurately, timely and with end-to-end traceability, means banks have more on their plate.

“This is where fintech and regulatory technology solutions like Hexanika play an important role,” Pandit says. “These solutions enable financial institutions to adopt new technology like AI, machine learning, [robotic process automation] and others to automate and add value to their business processes, data security and management, customer analytics, wealth management insights and several other areas.”

David O’Connell, a senior analyst with Boston-based Aite Group, contends that AI is somewhat more available for financial institutions of all sizes, since so many fintechs, such as the core providers and smaller players, offer options to community bank customers. “AI applications are being robustly offered by vendors,” O’Connell says. “Being able to do it, not on-premise, lowering the barriers to entry … smaller banks will be able to use those capabilities as easily as a $30 billion bank.” In fact, he says smaller banks are using this technology in process automation and fraud detection.

And then there are the potential cost savings that machine learning and AI may offer banks. A report from Capgemini’s Digital Transformation Institute predicts the financial sector could add $512 billion to its global revenues by 2020 and increase costs savings by 10% to 25% thanks to intelligent automation.

“In 2017, financial firms quietly introduced a range of practical machines that think,” the PwC report says. “Some banks added AI surveillance tools to thwart financial crime, while others deployed machine learning for tax planning. Wealth managers can now offer automated investing advice across multiple channels, and many insurers now use automated underwriting tools in their daily decision-making.”

A recent report from Business Insider Intelligence that reviewed important AI applications across banks found that front- and middle-office AI applications offer the greatest cost savings opportunities. It also found that banks are leveraging AI on the front end to improve customer authentication, enhance chatbots and voice assistants, and provide personalized insights and recommendations to customers.

“AI is also being implemented by banks within middle-office functions to detect and prevent payments fraud and to improve processes for [AML] and know-your-customer (KYC) regulatory checks,” the report disclosed. “The winning strategies employed by banks that are undergoing an AI-enabled transformation reveal how to best capture the opportunity.”

Collaboration, collaboration, collaboration

AI and machine learning are areas where banks and fintechs are working together to create customer-focused products that simplify existing processes and save time, money and resources, Pandit says. Hexanika has communicated with several small financial institutions, and the technology developer even built its own AI-driven custom development platform to support community banks in their AI implementation efforts.

Butterfield believes that banks like First National Bank of Omaha “are becoming easier to work with” as AI technology developers continue to emerge. While many bankers were previously fixated on compliance, Butterfield says more banks are taking a “belt and suspenders approach” to their efforts, working with core providers and emerging fintechs to create better applications.

“Everyone talks about regulations. ‘We can’t do this because of this or that,’” he says. “We act like every idea has to be taken to scale, when really we need to adopt a culture of experimentation.”

“We act like every idea has to be taken to scale, when really we need to adopt a culture of experimentation.”
—Marc Butterfield, First National Bank of Omaha

After recently authoring a report on the use of AI in lending, O’Connell sees AI as bringing “more wisdom to the process of banking … whether it’s looking at borrowing or a UCC filing. The next best action will be really important.”

He believes AI can be especially useful in lending, where there is “too much clutter in the loan life cycle.” In talking to bankers, O’Connell says, they are interested in using AI to “get to the edges [of lending] and improve where it is not automated enough.” Sykes sees AI leveraging nontraditional data sources to improve credit models and a banker’s ability to evaluate the likelihood of repayment.

He believes AI has an array of other use cases, such as with emerging robo-advising services technology, which has expanded the portfolio management market from just high net worth clients to those with smaller, traditionally costly investment portfolios. Businesses are implementing automation with voice and biometric customer authentication, reducing the time it takes for a representative to validate a customer’s identity. He says this presents huge time savings in call centers, where it can take 45 seconds or more for an agent to validate. And chatbots and voice assistants fueled by AI have emerged and are improving over time.

These technologies contribute to more effective fraud and threat detection, using the breadth of data and sources at the heart of a bank’s threat detection or risk framework to improve its ability to identify bad actors.

And, generally, machine learning will continue to integrate an ever-increasing number of data sources to further enable personalized service, advice and experiences for businesses of all kinds.

“AI and machine learning can enable institutions to automate and consolidate their processes, meaning manual and repetitive tasks can be performed by machines, freeing up resources to work on critical activities,” Pandit says. “[But given] the rising data requirements and frequent changes to compliance processes, community banks especially are struggling given the limited budget and resources at their disposal.”

Speare says few banks with less than $5 billion in assets are experimenting with AI. Credit card companies Visa and Mastercard are particularly interested. But while AI may take time to make its way into smaller community banks, the potential is great.

“There’s lots of interest in how AI can solve efficiency problems for banks, as well as retaining customer relationships,” Speare says. “The cost of a customer leaving is much greater than acquiring a new one.”

The perennial challenge of finding good people

David O’Connell

Artificial intelligence has the potential to bring many benefits to banks, from back-office efficiencies to more responsive customer service. But here’s the rub: There are just not enough qualified AI experts to keep up with the demand.

Seven of the largest U.S. commercial banks have prioritized technological advancement with investments in AI applications to better service their customers, improve performance and increase revenue, according to PwC. But with tech giants offering higher salaries, stock options and other perks, even megabanks are finding it difficult to recruit top AI talent.

“Despite the fact that banks are having a difficult time attracting data scientists for their AI road maps” says David O’Connell, a senior analyst with Boston-based Aite Group. “Everybody is interested in it. There’s a real appetite out there.”


Karen Epper Hoffman is a writer in Washington state.

Top