••• SPONSORED SECTION •••
Once relegated to performing traditionally mundane tasks, AI is increasingly showing its sophistication and potential through payment system innovations.
In today’s increasingly digital world, businesses everywhere are embracing digital innovation to stay competitive. Central to this movement is the often-referenced, and seldom-understood, phenomenon of Artificial Intelligence (AI) – and machine learning (ML), specifically. While these terms are sometimes used interchangeably, there is an important distinction, says Charlotte-based Doug McKinley, PNC senior vice president and head of treasury management innovation. “AI is the science of training machines to perform tasks that would normally be performed by people,” he says. “Machine learning is a form of AI. It entails providing machines with large amounts of data so they can identify patterns and draw conclusions, in the same way humans do, but better and faster.”
With massive volumes of data being generated by a growing range of networked devices – and backed by ongoing advances in computing power – the use of AI and ML in business continues to accelerate. The financial services industry is among those harnessing the power of AI and ML to help protect, support and empower customers. Reducing exposure to fraud, executing faster payments, and providing data-driven business insights are among the capabilities AI can help enable.
While AI and ML drive automation, optimization and intelligence throughout the financial services ecosystem, the adoption and integration of AI-based engines and ML algorithms is perhaps most visible in the realm of treasury management solutions. “While technology continues to be the main disruptor in banking, AI takes on even greater relevance for companies when we think about the continued shift of paper-based payment systems to digital – a change that accelerated significantly in 2020 as many companies transitioned to a virtual working environment,” says Charlotte-based Chris Ward, PNC executive vice president and head of treasury management digital and innovation.
Ward and McKinley both have contributed to the delivery and integration of AI- and ML-enabled treasury management technologies – beginning in the early days of robotic process automation more than a decade ago. Today, machine learning is no longer a curiosity or novelty, says Ward. As reflected in the following usage examples, it’s a business imperative.
AN IMPORTANT LINE OF DEFENSE
Because of ML’s propensity for recognizing patterns in data and behavior, it is commonly employed to help identify fraudulent transactions and activities. The significance of this capability cannot be overstated; the 2020 Association for Financial Professionals Payments Fraud & Control Survey, which examines fraud attacks on B2B transactions, found that 81% of organizations experienced actual or
attempted payments fraud in 2019. And that was before the pandemic, which created an even more friendly environment for fraudsters to exploit the vulnerability, uncertainty and change that many companies and employees experienced.
ENABLING SAFE REAL-TIME PAYMENTS
The convergence of mobile technology and digital commerce has ushered in real-time payment innovations, including the Real-Time Payments Network (RTP), which can allow participants to send and receive funds immediately at any time. AI- and ML-enabled technologies are crucial for monitoring and analyzing transactions in the fractions of a second necessary to facilitate real-time payments.
AUTOMATING FOR EFFICIENCY
As the pace of business increases and companies continue to embrace efficiencies, automating the receivables process is one example of a function that ML-enabled tools have helped deliver. PNC Receivables Automation, for example, draws upon AI-based engines and machine learning algorithms to match incoming payments and associated remittance of open invoices.
LOOKING TO THE FUTURE WITH CASH FORECASTING
Currently in development at PNC is an ML-enabled cash forecasting solution, designed to help treasurers see their future cash balances.
“Traditional cash forecasting is largely dependent on spreadsheets and manual data input, so it can be time-consuming and prone to error,” says Ward. “A treasurer can spend a lot of time building a model, only to arrive at a quasi-accurate forecast – and with little time to actually address learnings from the forecast. This solution aims to simplify the process and allow the treasurer to spend more time using the cash forecast and less time building it.”
Ward and McKinley are quick to point out that the cash forecasting solution won’t be the final example of an ML-enabled tool in the treasury management toolkit. As Ward puts it, “We have barely scratched the surface.”
