This website uses cookies

Read our Privacy policy and Terms of use for more information.

For a gig worker waiting to get paid or a small business managing cash flow, a few days can make a big difference. But even today, with all our advances in technology, moving money in seconds requires more than a faster payment network. Banks also need technology and operations that can support transactions around the clock.

That is one of the challenges facing Avik Chatterjee, a managing vice president and divisional chief technology officer for payments at Capital One. A payments industry veteran who previously worked at PayPal and Venmo, Chatterjee oversees technology that supports money movement across Capital One’s consumer, small-business and commercial businesses.

In a recent conversation with the Financial News Network, Chatterjee and I discussed how changing expectations around getting paid are pushing banks to modernize their infrastructure. We also talked about where and how Capital One is using AI and how real-time payments are already speeding up funding for auto dealers.

This interview has been edited for length and clarity.

What is changing in payments that will meaningfully affect how people and businesses move money?

Real-time payments play a huge role in how the underlying financial infrastructure is changing. Think about the new ways people are getting paid. For Uber drivers and content creators, the cycle between creating something of value and getting reimbursed for it is getting shorter and shorter.

Those units of value have to move over the new payment rails evolving across the ecosystem. The expectation is that money movement will keep pace with our lives, especially in the gig economy.

Avik Chatrerjee

We’re also seeing that in business-to-business interactions. The cycle between providing a service and getting reimbursed is shrinking. That is especially important for small businesses, where cash flow matters so much. Even a few days’ difference in money movement could make or break a business.

What does that shift require of banks’ technology, and how does Capital One’s move to the cloud help?

The software we build runs on cloud infrastructure rather than in our own data centers. That gives us the flexibility to deploy infrastructure across multiple geographic locations, so we can shift traffic if one of our operational units is down.

As the industry moves toward real-time payments through networks such as The Clearing House’s RTP network and FedNow, the question becomes: How do we make our infrastructure always on and always available, in line with those networks’ expectations?

With older operating models, banks built in downtime for maintenance and end-of-day processes. When a primary data center had a problem, shifting traffic to a backup could take time.

In a real-time payments ecosystem, it isn’t sufficient for one financial institution to be up. If the others are down, it cannot send money to them. Banks have to think about what 24/7 operations look like.

Looking back at Capital One’s journey to modernize our technology and deploy software in the cloud, we're in a strong position to embrace real-time payments.

What are the biggest problems your payments team is trying to solve over the next year?

One is removing roadblocks to 24/7 operations. While we have been building modern software and deploying it on cloud-native architectures, Capital One has also grown through acquisitions. We have inherited some legacy mainframe applications that can get in the way of where we want to be.

We are continuously working to bring those parts of the process onto infrastructure we are comfortable building, owning and maintaining.

❝

We have inherited some legacy mainframe applications that can get in the way of where we want to be

Avik Chatterjee, managing vice president and divisional chief technology officer for payments at Capital One

From a customer-facing perspective, we’re looking at how to enable more real-time money movement. We already supported The Clearing House’s real-time payments network, and having FedNow as well gives us additional resilience. During a recent TCH network outage, we received payments through FedNow. That redundancy is important.

We are also investing in our debit capabilities, where AI helps us innovate responsibly. When you’re dealing with people’s money, trust and security are the foundations of the business.

How do you balance faster payments with fraud prevention?

One challenge with real-time payments, particularly payments with a degree of finality, is evaluating risk and compliance considerations while deciding on a transaction.

With debit cards, the end-to-end window is sometimes two to two and a half seconds. Within that time, we are continuously improving AI models that can evaluate risk, reduce false positives and improve transaction approval rates.

Those are also areas we would explore as we scale real-time payments through RTP and FedNow.

Where else is AI improving payments and banking at Capital One?

There are two aspects: using AI to improve how we build software, and using AI to help the customers we serve.

Internally, we have seen a significant acceleration in how quickly we can develop software, build products and features, and get them in front of customers. We are committed to giving our developers access to the latest tools, frameworks and frontier models, and we are seeing a tangible increase in the pace at which they work.

On the customer-facing side, fraud mitigation is one area. In my world, that includes debit transactions and incorporating those capabilities as we expand real-time payments.

We also have AI innovations that have accelerated back-office operations, specifically around customer service. Agents on calls have to make decisions, and AI-enabled workflows can help them work faster and address customers’ needs more quickly.

In retail banking, we’re also exploring ways AI could reduce the work involved in managing someone’s financial life. One example is automatic savings: Models could decide to move money from checking to savings based on the cash-flow patterns we can see.

❝

Agents on calls have to make decisions, and AI-enabled workflows can help them work faster and address customers’ needs more quickly.

Avik Chatterjee

(Reporter’s note: The bank recently rolled out Autopilot, a product that uses AI and machine learning to automate small transfers into savings. It reminded me of what startup Rivo is doing)

I recently covered a startup working on something similar to the automatic savings you described. How is Capital One approaching that?

Chatterjee: All of our lives are busy, and we’re exploring ways AI could help customers make financial decisions. One example is automatic savings, where models could decide to move money from checking to savings based on the cash-flow patterns we can see. It’s another way we can make customers’ lives easier without the cognitive load of managing their finances.

If AI agents start making more purchases on behalf of consumers and businesses, how will fraud prevention need to evolve?

As an industry, it’s a very nascent space. Fraud models take time to evolve as we see these new vectors emerge, so we are keeping an eye on it.

I think the industry’s fraud models will evolve to make those experiences more seamless for customers over time.

If we spoke again a year from now, what would you expect to be different about Capital One’s payments business?

There are two parts to it. The first is supporting networks such as FedNow and RTP on our platforms in a well-managed way, with the trust, security and operations those networks require.

The second is expanding those offerings so customers can access them directly through our digital channels. That is an interesting area for us to evolve.

One example of how we already use real-time payments is auto loan disbursements to dealers. Capital One has a large network of car dealers. When a customer completes the process of buying a car through our ecosystem, we make sure the dealer gets funded in a matter of seconds instead of days.

That has been one of our initial uses of RTP to make the technology tangible for customers. We will continue exploring how to expand access to those capabilities.

Are stablecoins something you are considering?

We’re keeping our eyes and ears open. We are always looking at whether these innovations can help us serve customers better and create interesting experiences. From that perspective, we are keeping an eye on them.

What is holding instant bank payments back from becoming an everyday way to pay?

Payments are a two-sided network. Merchants and other institutions also need to accept those payment methods.

One area where we are seeing early adoption is government-to-consumer payments. I think FedNow will play an increasingly important role in payouts related to natural disasters, when people need the money most.

I recently received an insurance payout for my car through a push-to-debit experience. More companies are starting to send funds immediately to consumers through debit networks. We also need to explore how to expand RTP and FedNow into those kinds of experiences.

Multiple payment rails are now available to businesses and agencies to fulfill that promise. In my case, I needed to buy a new car. Having access to the insurance payout immediately made those cash-flow decisions easier.

Reply

Avatar

or to participate

KEEP READING


VIEW MORE