AI in Accounts Receivable in 2024

Time to get serious about AI. The technology has bright promise in all sorts of applications.

Given the exponential growth of adoption for artificial intelligence over the last 12 months, it’s only natural that leaders across industries are eager to map out how they might use (or continue to use) AI in 2024.

Though the electronic bill payment and presentment (EBPP) industry must exercise particular prudence in how it approaches any new tech, leaders in the space are as anxious as anyone to put this exciting new technology through its paces and discover how it can benefit the payments sector. Based on my 21 years working in payments, I think there are three areas where we’ll almost certainly see applications (or continued applications) of AI in 2024: reconciliation, underwriting, and fraud detection.

Reconciliation

Reconciliation and invoice-matching are obviously core undertakings in accounts receivable. They are both hugely important and hugely time-consuming. AI’s greatest strength today is automating repetitive, labor-intensive tasks just like these, reducing to mere seconds what otherwise would be several person-hours. Reconciliation is ripe for this kind of overhaul. Furthermore, using artificial intelligence to automate repetitive tasks like these reduces the risk of human error that can result from fatigue or even simple boredom, and frees up time for humans to work on tasks that require greater discernment.

In fact, I’d argue that one of the main reasons reconciliation is an excellent candidate for artificial intelligence is the fact that it doesn’t actually require much, well, intelligence. Reconciliation gets a bad rap for being tedious, but it isn’t actually very difficult. The actual processes of reconciliation are relatively straightforward, if repetitive.

Where the issues crop up, and where human intelligence remains crucial, is in identifying discrepancies and probing those differences to iron out errors. So while I’m optimistic when it comes to leveraging AI in reconciliation, I expect that human oversight will continue to be necessary at least through 2024, if not quite a bit longer.

Underwriting

AI’s capacity for lightning-fast data analysis could be extremely valuable in some of the fundamental elements of underwriting, a process that traditionally sees underwriters individually parsing information to establish a solid risk assessment. Most straightforwardly, artificial intelligence can take on a slew of risk-assessment tasks. Its pattern-recognition capacities make it a great candidate for extrapolating trends from data concerning payment behavior and history.

This doesn’t eliminate the need for a human touch. On the contrary, AI’s pattern recognition would free up time and space for people to concern themselves with that which deviates from the pattern. Some anomalies in historical patterns are evidence that suggests risky behavior, but some are simply exceptional events that all people encounter at one point or another.

Human underwriters will be able to focus on these events and apply their common sense and situational understanding to better understand their significance—something AI is not capable of doing.

Beyond that, AI also has the capacity to add nuance and dimension to the process of credit scoring. Where today, we rely on analyzing historical data and behavior to determine a credit score, AI will be able to take into account not just historical, but also real-time data—including up-to-the-minute behaviors and market status.

This adds dimension (and accuracy) to any credit score, but it also benefits those with little or no credit history. By including live data, the depth of information available to predict payment behavior is enhanced. AI can make all underwriting activities more dynamic and agile, empowering real-time adaptation to changing financial environments.

Fraud Detection

Pattern recognition will probably always be computing’s greatest strength, and the algorithmic capacity for recognizing patterns has more than outpaced that of a human. This makes AI a potentially invaluable tool for fraud detection, as it is able pick up on even the most minute anomalies among enormous swaths of transaction data—and do it in real time.

Given how self-evidently logical it is to leverage AI this way, it’s no surprise that, in fact, this kind of fraud-detection AI already exists. I anticipate seeing more and more tools crop up, and their use proliferate exponentially, in the coming year.

I think anyone working in the electronic bill-payment and processing space would agree that reliable, robust fraud protection is intrinsic to the integrity of any accounts-receivable practice. That’s why I’d say it behooves us as a profession to consider any and all tools that enhance it. It’s part of the higher standard we’re held to across the industry, and protecting our customers has to remain our priority.

In my two decades’ working in the payments space, I’ve seen waves of new tools introduced into the industry—with varying degrees of success. I’m of the opinion that, in general, all industries, including EBPP, should keep an open mind and a healthy optimism when it comes to novel technology. So I look forward to seeing the innovative ways accounts-receivable leaders will leverage AI to better serve customers.

That said, no matter how well we’re able to automate many of the tasks and operations involved in EBPP, there will always be a need for smart, dedicated people. No matter how much it may seem like accounts receivable is a hard science, there is an undeniably human element to what we do.

Payments support access to the goods and services people rely on every day. The sector will always require humans not only to monitor AI models and refine them in compliance with evolving regulatory standards, but to maintain humanity in what is a human industry.

—Sara Faied Phelps is vice president, payment operations, at InvoiceCloud