Collaboration—and reams of data—are key to making AI smarter
With the help of two AI tech firms, two disparate credit unions here are hoping to redefine how members use chatbots when banking, including executing transactions.
“We wanted to be an early adopter of this delivery channel and be prepared when these devices become ubiquitous in digital banking, rather than playing catch-up,” said Jim McCarthy CEO for the $116 million Portland, Ore.-based Trailhead Credit Union. The credit union supports 9,037 members at three branch locations and has 8,400 active online banking members and 3,100 active mobile members.
“The security of members’ financial information is paramount and it’s something we have had questions about and expect members will, too,” said McCarthy. “The chatbot does not have access to members’ financial or account data and cannot be used to answer questions that are account-specific. Rather, the information that the chatbot provides mirrors the information that is readily available on our website.”
McCarthy explained that the “chatbot” technology was made possible by the Orlando, Fla.-based Abe AI, which offers services to the Northwest Credit Union Association’s (NWCUA) member credit unions via Strategic Link, Inc.
“With our platform, credit unions can create AI-powered chatbots that simplify banking experiences through Amazon Alexa, Facebook Messenger, SMS and Google Home,” said Juan Romera, head of business development for Abe AI. “For credit unions that don’t want to build their own AI-powered chatbots, we’ve developed money management, rewards and customer support chatbots that are market-ready.”
Trailhead CU members can access the bot via Google Home by saying, “Ok Google, talk to Trailhead Credit Union.” For Amazon Alexa/Echo users, members have to open the Alexa app on their phone and search for the skill “Trailhead Credit Union” and select “enable.” Once it’s added to the app, members say, “Alexa, open Trailhead Credit Union” to start a conversation.
“Our two-person IT department was kept updated from kick-off to launch, but Abe AI managed the entire process from a technology perspective and the integration onto each of the channels,” noted McCarthy.
Trailhead CU launched the new solution in January, which McCarthy said doesn’t require a user login. While new user data is streaming in daily, he explained that members are asking about a variety of subjects, including branch information, routing number, contact information and help with navigating Trailhead’s services, such as: “Ok Google, how do I set up direct deposit?”
“Observing the type of questions that are being asked is very important in furthering its development. This is such a new technology for digital banking, that it’s most useful to see how our members want to use it, first and foremost,” said McCarthy. “As we learn more about what members’ needs are on these platforms, we will work with Abe AI in a phased approach to include additional utilities they may offer.”
Taking chatbot a few steps further down the line
Having access to general credit union information via voice is beneficial to members and cuts down on employee overhead. John Best, CEO of the Tampa, Fla.-based Best Innovation Group (BIG), however, is pushing chatbot capabilities further: executing transactions.
Late in 2017, Best announced the certification of its Amazon Alexa voice banking solution, the Financial Interaction Voice Experience (FIVE) banking skill. The $456 million Oak Ridge, Tenn.-based Enrichment Federal Credit Union became a credit union market leader offering this single sign on transaction-based functions, such as transferring funds between accounts and executing loan payments.
“This is a real substantive application of artificial intelligence,” said Enrichment FCU CEO Craig Peters. “We don’t look at this as a credit union function. Alexa, on its own, can control your lights and your security system. These types of applications are not going away and will continue to grow. We are injecting ourselves into the same process and I don’t see that going away anytime soon, either.”
Enrichment CIO John Merritt explained that the credit union supports more than 46,000 members, 17,500 of whom are active online bankers. After two months of use, there were 130 unique members using the solution as well as 390 sessions (where a user has initiated a new session with Alexa (to re-authenticate). Members have also logged more than 1,660 “utterances,” which are requests, such as an inquiry, transfer or request for other information.
“We didn’t expect a flood of membership to adopt this off the bat,” said Merritt. “The biggest thing for us was to create buzz and be the first one in our market.”
Peters explained that a number of physically challenged members, including those who are blind, are among early adopters. He added that the credit union bought Amazon Dots for all 115 employees and also offers a free Amazon Dot for new members agreeing to use the solution. The rate of members making an information inquiry versus a transaction function is nearly fifty-fifty, he added.
“I think you will see as much traction on this as we saw on remote deposit capture,” said Peters.
For Best, who is currently working on related AI Google Home and Facebook chat platforms, the success of AI bots in the credit union space will be based on CUs sharing AI-triggered and aggregated data. The reason Siri and Alexa continue to get smarter is by the volume of requests they receive on a daily basis, he said.
“A single credit union is not going to get that kind of scale of data. It may have thousands of requests per month, as opposed to billions. What we are doing with FIVE is the same as this analogy: If the water goes up in the harbor, all the boats rise with it,” said Best who added that there are a number of credit unions that soon will begin using FIVE, including the Spokane, Wash.-based Numerica Credit Union, which serves 137,000 members.
Best explained that the goal of FIVE is to not have a separate process for Google Home and Alexa and the vast majority of other AI assistants, such as Abe AI. Rather, he wants to create a single platform that communicates with the multitudes of credit unions’ AIs so the intelligence is shared and reused. Parity between channels, he says, is critical to success.
“We are working on a collaborative AI channel to combine numerous credit unions AI data and teach on that scale,” said Best. “So scale and time are working in our favor and we can make one AI that is smarter over time for the credit union industry. We want to turn this into a true personal financial assistant as opposed to an in-house IVR (interactive voice response).”