Capital One AI VP discusses AI assistant Eno

Eno from Capital One is an AI assistant that can give customers real-time banking updates and alerts to possible fraud attempts. In this Q&A, Capital One's VP of conversational AI goes over the basics of Eno.

Now that smartphones are universal and AI technologies are embedded in consumer and enterprise software, people have come to expect real-time notifications and deep levels of personalization, whether it's from their video streaming or financial apps.

Eno, from Capital One, is an automated personality that can provide Capital One customers with AI-enhanced levels of insights, personalization and security.

Rolled out over the last couple of years, the Capital One AI agent can alert customers about purchases that seem abnormal, upcoming bill payments and other credit card issues, offering actionable notifications and tips across a number of channels that include SMS, browser plugins and push notifications.

According to Ken Dodelin, vice president of Capital One's conversational AI product development, Capital One intends Eno to make finances a little easier, safer and more interactive for customers.

In this Q&A, Dodelin describes the uses of the Capital One AI assistant, as well as its development and its customer reception.

Editor's note: This interview has been edited for brevity and clarity.

What are some insights the Capital One AI assistant, Eno, can provide to customers?

Ken Dodelin: So, money is hard, and it can be complex; it can be stressful. We come at this with a 'Wouldn't it be great if everybody had a really smart AI that knew them and their needs, and was working on their behalf in the background.'

Capital One AI assistant Eno screenshot
Capital One's Eno provides AI-driven fraud notifications and insights.

[For example,] Eno will reach out and let you know when you've been double charged -- right, two identical charges from the same merchant show up -- maybe that was a mistake, and Eno will reach out … and help you resolve it if it was a mistake. Another example is maybe you just left an exorbitantly large tip, and Eno might note that that is atypical.

Another is you have an automated subscription that's about to renew -- maybe you did it a year ago and you weren't aware -- and Eno can detect when that's about to happen and give you a chance to take action on that.

How have customers reacted to Eno?

Dodelin: They love the notifications -- the positive feedback has exceeded our expectations. People just love the convenience of a quick text to get information and then get back to living.

That's really what this is all about, right? Do people wake up in the morning thinking 'I can't wait to start banking today?' Well, it's more that they are banking to enable other things they want to do in their life, and having an AI working on their behalf in the background can help them spend less time banking and more time living.

What data is used to power Eno, and how are the Capital One AI algorithms developed?

Dodelin: OK, so there are a couple parts to that question.

First, I think the data is the same. Maybe the availability of it internally has evolved a bit. The, you know, customer bank relationship for many decades has been, 'Here's the information; you figure it out.'

Instead of training humans to interact with the machine, we're training the machine to interact with the humans.
Ken DodelinVice president of conversational AI product development, Capital One

I think what we're seeing is that we now have an opportunity to evolve that relationship, to add a little more value by providing insights and analysis and quick actions to be taken on behalf of the user and bringing them to the customer in the channel of their choice.

The burden of all the analysis and the work can be shifted a bit away from the customer, while leaving the customer in complete control.

As far as the algorithms [go], we've chosen to build them in-house. We built a natural language processing solution that learned how to speak 'bank' with our customers, and that's been a great exploration and a great tool for us.

Instead of training humans to interact with the machine, we're training the machine to interact with the humans and to understand how they speak versus how we understand.

What are some of the new and upcoming developments with Eno?

Dodelin: I think you're going to see more of Eno becoming more proactive -- anticipating customer needs and getting ahead of them so that people don't have to wait; they can just react and make decisions on things even before they may become a problem.

Ken Dodelin, vice president of Capital One's conversational AI product developmentKen Dodelin

I think we're going to continue to invest in intelligence that can understand more and more types of customers and more and more of the issues they bring to Eno. We will also be threading Eno through more of our digital experiences, increasing its presence in the places it is already available, and possibly going out to other ones.

With the advent of machine learning, and data availability and so forth, we are now positioned, I think, across many industries to provide Augmented intelligence to people. And so, the analysis of information that previously was left to the customer to do can now be automated and presented in a smart way that allows people to digest information efficiently and take action, which is a much more efficient process than leaving everything for the human to do.

Dig Deeper on AI technologies