J.P. Morgan Payments: AI Transforms Payments’ Tony Starks into Data-Driven Iron Men

Artificial intelligence is helping payments executives become high-powered problem solvers.

“What makes Tony Stark Iron Man is his exoskeleton and AI-powered enhancement technology,” Tony Wimmer, head of data and analytics at J.P. Morgan Payments, told PYMNTS’ Karen Webster.

AI is giving payments professionals, he said, a bit of their own high-powered enhancements in the mix. 

At a high level, he said, the initial impact of AI is on human decision-making by delivering the right insights at the right time. 

Those insights, said Wimmer, can help a company’s treasury, merchants, issuers and acquirers benefit from the opportunity to use AI and data-driven insights to optimize — and repair — payments by increasing conversion rates.

Wimmer noted that as card transactions go from merchants to acquirers to networks and issuers, the inherent transaction messages have many data elements and can be delivered in literally a million different permutations. The ecosystem has milliseconds to determine whether or not a transaction may be fraudulent. Every card issuer has its own fraud scoring engine, and the payments realm is constantly adding new data fields, he said, so complexity reigns. “Most of the fraud decisions are false positives,” he said, “and the goal is to try and reduce that.”

AI can help merchants grow and solidify customer loyalty as they understand their customers’ buying behaviors, he said. Forward-thinking merchants can use transaction data to determine their best customer segments, what loyalty offers might be most useful and when, and all of it done in a privacy-compliant manner. 

Consumers want relevant communications from merchants. Armed with real-time insights, merchants can be proactive in crafting those communications with customers rather than simply asking for emails and mobile phone numbers so that they can reach out in hopes of prodding more sales. 

“AI,” he said, “can find patterns that are invisible to the human eye and bring attention to critical issues so that humans can focus on what they do best: find creative solutions to complex problems.”

But to get there, he said, the very nature of organizations has to change and get the right technology and data in place to get those insights in the first place — all of which requires significant investments of time and resources, including humans, at the intersection of business and data science.

“I think we’ve established by now,” he said, a bit tongue in cheek, “there is a lot of data in payments. But historically, the data has not been organized well enough for AI to actually work.”  

That’s especially true when payments are moving at scale — and J.P. Morgan Payments, he said as an example — moves $10 trillion worth of payments volume daily. Collecting the most basic details of a transaction can be a challenge. And it’s essential that AI applications have “clean” data with which to work. Software-as-a-Service solutions from providers such as J.P. Morgan Payments can help clients harness that data to turn signals into insights.

Making Payments Safer

“Embedding those signals back into the transaction,” observed Wimmer, “also makes the payment ecosystem safer — and more efficient than ever.” Fraud scoring, after all, is not a one-size-fits-all model. AI-underpinned account validation enhances the transaction’s safety, which can help avoid ACH payments being returned because beneficiary account information is outdated or numbers have been erroneously coded.

“About 15% to 20% of ACH returns of this nature can be prevented,” Wimmer said. “We have machine learning models that monitor hundreds of millions of accounts across the payments world — and we can estimate the probability of success based on the last time we’ve seen successful payments to a given account. These are account pattern recognition models.”

The fraud models, he said, can also create “beneficiary trust scores” that offer up a risk-adjusted number that helps an organization decide whether they want to let a transaction through or not.

As time goes on, he said, the application of AI to payments will become a strategic differentiator for a company’s treasury, merchants and acquirers born of necessity as these entities improve the delivery of their services. In one example, J.P. Morgan Payments has invested in a Software as a Service solution for forecasting cash flow that can help treasurers more accurately predict inflows and outflows — essential to managing a company’s liquidity.  

In the meantime, he said, fears that the machines will displace humans are misplaced.

“There’s a long way to go before there’s a futuristic version of AI where machines think and make decisions … humans will be around for quite a while,” he said. “And the more that we can write software that has payments data at the heart of it to help humans, the better payments will get.”