Smart Agents Are AI’s DNA — Here’s Why

Inspiration comes from anywhere and everywhere, and that goes for digital innovation in payments and commerce. Just ask Akli Adjaoute, CEO of Brighterion (a Mastercard company), who recently spoke with PYMNTS’ Karen Webster about the rise of smart agents, a form of artificial intelligence designed to bring more fraud prevention and customer service to financial institutions and other digital players.

As Adjaoute told Webster during a recent interview centered around the launch of a new PYMNTS-Brighterion report entitled “The AI Innovation Playbook: Moving Toward a Future of Smart Agent Adoption,” his background in medical science and his service in the French military resulted in his interest and work on smart agents.

Simply put, smart agents is technology capable of unsupervised learning that can adapt to complex ecosystems and can be assigned to perform particular functions. They can, for instance, monitor the behavior of a specific cardholder, and join with other smart agents to provide a collective intelligence that can be used to detect fraud patterns, among other tasks important to the FIs that deploy the technology.

Smart Agent Inspiration

Adjaoute’s experience with medicine and military life provided vital lessons that now inform his work with smart agents.

First, people may share similar DNA and suffer from the same diseases, but effective treatment requires taking into account how one sick individual is different from other people – call that the aspect of personalization, a big part of the smart agent value proposition, at least in his telling. Second, his work in the military taught Adjaoute about dealing with “mission-critical” issues with “optimal” solutions that fix important problems – smart agents hold the promise of making consumer life (along with FI operations) safer, more efficient and more lucrative.

“The key here is personalization,” he said of smart agents and the work for which they are designed. “You have to know (consumers) by their unique and differing behaviors. Other tech, such as business rules and data mining, cannot provide such personalization for financial institutions.”

And that’s not all.

A successful smart agent deployment also requires the technology to be adoptable – trying to fight, say, malware attacks one by one is an impossible, frustrating path, so it’s better to let AI-enabled technology take the fruits of its lessons and apply them to future threats. And, of course, self-learning is another top capability for smart agents.

AI Gaps

The thing about artificial intelligence in 2019, most notably when it comes to banks and credit unions, is that everything seems to be talking about it, but few are really doing it. More common than artificial intelligence is the use of machine learning (that is, supervised learning) technology and techniques, at least according to PYMNTS research. That stems in part from confusion about the differences between artificial intelligence and machine learning – a confusion that Brighterion, among other companies, is working to overcome.

That said, the interest in AI and in smart agents is there, and significant.

The PYMNTS-Brighterion research shows that 41.1 percent of commercial banks are “very” or “extremely” interested in adopting smart agents. Not only that, but 45 percent of decision makers working in fraud detection are interested in adopting smart agents. And 81.8 percent of financial institutions with assets exceeding $100 billion are interested in using smart agents to enhance their fights against fraud.

False Positives

The appeal of smart agents, according to Adjaoute, comes not only from personalization and those other factors, but also their ability to build “real-time profiles” of customers, a skill that enables the financial institutions to keep track of life changes for its customers (new job, new marriage) while also protecting against fraud. And the smartness of those smart agents also can work – individually and collectively – to severely reduce false positives, those annoying sources of friction that keep a legitimate consumer from completing a transaction because he or she has wrongly been flagged as a fraud risk.

That, in turns, allows the FI, via its smart agents, to adopt a proactive view – a view that can bring benefits to customers, including by reducing friction. “The key is robust collective intelligence,” Adjaoute told Webster. The work of smart agents “can detect things that might happen to you, so you don’t need to look in the rearview mirror and hope that the next time it doesn’t happen.” That applies even to such fraud threats as money laundering, something the collective intelligence of the smart agents can protect against.

So what are keeping those FIs from deploying more smart agents?

According to the new PYMNTS research report, the reason has more to do with FIs’ own limitations, not the technology of smart agents. Sure, 58.1 percent of FI decision makers surveyed for the report said smart agent benefits are intangible. But other hindrances include a lack of employees with the skillsets to handle the technologies (50.0 percent) and the belief that the technologies are too complicated (35.1 percent). Neither issue pertains to the abilities and limitations of AI and ML, but rather those of the FIs using such solutions.

But when it comes to the future of smart agents, Adjaoute is optimistic – and in a way that goes beyond him just promoting his company. He comes off a true believer, and when you hear him discuss this specific part of the digital world, you detect sincerity, the tone of a man whose life has provided ample inspiration for this innovative work. “All my life, I cared about mission-critical AI, and the most important thing for me was being able to offer something of value,” he said.