Like virtually each different financial institution, Capital One used to have a fundamental SMS-based fraud alert system, asking prospects if uncommon exercise that was detected was real. But in about 15% of instances, folks didn’t textual content again a easy sure or no.
As Ken Dodelin, VP of Conversational AI Products at Capital One, defined in dialog with VentureBeat CEO Matt Marshall at Transform 2020, a very good variety of folks would write again different explanations — issues like “Yeah, that was me” or “Yep. [thumbs-up emoji]” or “Yes, that was last weekend when I was in Philadelphia visiting my sister.”
“We didn’t understand any of it [then],” stated Dodelin. “And so that kind of hit the lightbulb for us — wouldn’t it be great if we could understand natural language coming back to us from customers?”
Since then, the imaginative and prescient of constructing an AI assistant that takes complexity out of cash for Capital One prospects, and makes cash administration simpler, has been relentless. The deliberately non-gender-specific-named ENO requires no opt-in. Through SMS or push notification, ENO will let you already know in the event you have been charged twice for a similar quantity inside a few seconds; it’d test if an exorbitantly massive tip you left was intentional, or if the decimal place was within the unsuitable spot. Or it may provide you with a warning that the free trial you signed up for (and clearly forgot about) is about to run out.
At the outset, the corporate relied on a third-party NLP/NLU answer. However, a profitable funding in constructing in-house expertise tipped the scales of the buy-or-build conundrum. “We brought in one of the folks that worked at IBM Watson and built the AI that won on Jeopardy,” stated Dodelin. “And that team was able to build a banking-specific NLU that outperformed the general one that we were relicensing.”
Capital One now can perceive 99% of buyer replies (versus 85%), presents quicker response occasions for confirmed fraud, and supplies a greater buyer expertise — as a result of prospects are understood.
Another essential funding for the corporate has been in a cloud native utility. Becoming an early cloud adopter has enabled the financial institution to each decrease prices and run its infrastructure extra effectively. And the compute energy to extract and analyze knowledge continues to result in higher outcomes.
“Building and releasing models isn’t the same as building and releasing widgets,” stated Dodelin. “There’s a lot of trial and error and learning that goes on. We have a tremendous opportunity to unleash the power of data and customer context into these experiences. And the more that that data becomes available in real time, the more we can train our models, the more context we can give them, the better we’ll be able to get to that one-to-one personalized experience.”