BofA Augments Internal EricaAssist with GenAI, Eyes Erica Virtual Assistant Next
By Steve Cocheo, Senior Executive Editor at The Financial Brand
Simple Subscribe
Subscribe Now!
Bank of America is rolling out GenAI functionality for EricaAssist, the version of its Erica AI that contact center employees use to solve customer problems and answer queries. It’s a new step for BofA, as the version of Erica that customers use within the bank’s mobile app to date has not incorporated GenAI to date.
Erica, the virtual assistant for customers, launched in mid-2018.
“We based it on the best AI available at the time,” recalls Tom Ellis, CIO and head of consumer technology and a BofA veteran. “We used open-source models to do that. And we’ve trained it over time on the basis of client interactions.”
EricaAssist followed in October 2020, based on learnings from interactions with customers.
Key point: As GenAI is being added to EricaAssist, Ellis says BofA is also in the midst of developing “Erica 2.0” on the customer-facing side and that GenAI will be incorporated into the new version.
The timing with GenAI implementation is intentional. “We want our contact center agents to have these new capabilities first because that way we’ve got a human in the loop,” explains Ellis. In that way, he says, “we can try out the new capabilities, but the agents are the ultimate decision makers.”
Need to Know:
- Customer-facing Erica hit 24.6 million users in the second quarter, up 23% year over year. There were 200 million Erica interactions in the quarter, up nearly 15% year over year. (This aspect of Erica serves consumers as well as other segments.)
- Erica has handled more than 3.6 billion interactions since its 2018 launch.
- More than 1.9 billion of those interactions have been proactive and personalized insights for customers.
- A lesson about virtual assistants: Ellis says at the beginning the bank’s developers thought customers would be using their voices to interact with Erica and put a lot of stress on that. “In reality,” he says, “they don’t do it that much. They prefer to text with Erica.”
How Bank of America is Adding GenAI to EricaAssist
One of the key goals of adding GenAI to the internal version of Erica is taking some of the “no joy” work out of the customer contact center job, according to Ashley Ross, head of consumer client experience and business transformation. The augmented EricaAssist is always on, represented by a widget on each representative’s computer screen. When a customer call comes in, the system automatically serves up the nature of the customer’s problem and key details about their relationship with BofA.
More than 18,000 human agents have EricaAssist and the bank is exploring other parts of the bank where the software can improve performance.
Contact center associates’ feedback has driven decisions on tasks that GenAI will undertake initially.
Example 1: Eliminating notetaking and drafting drudgery. One example is using GenAI to generate summaries of each customer conversation in real time. Previously agents had to handle that themselves, juggling the summaries while listening to customers and solving their problems. Agents can pull up the AI-produced summaries for revisions.
“This way, they aren’t distracted trying to do a lot of manual work while they’re on the phone with clients,” says Ross.
In addition, the GenAI can help direct the guidance it provides to the bank’s agent. Some customization of recommendations occurs as augmented EricaAssist analyzes the customer’s history with the bank.
A key role of EricaAssist, improved and hastened by GenAI, is finding guidance in context for agents. The bank says that this can now be provided in seconds.
Example #2: Pinpointing why card transactions were declined. Ross says an example of how this capability can help is customer queries about credit card transaction declines.
Before GenAI was added agents had to access multiple bank systems to learn more about the declined transaction itself and why it wasn’t approved. Such manual research has been drastically reduced.
“Card declines sound pretty simple,” says Ellis. “But there are upwards of about a thousand different reasons for a card decline.” He says that adding GenAI allows much of the process to happen in the background and more quickly than before. That saved time can be spent addressing other concerns the customer may have.
Overall, adding GenAI to EricaAssist is expected to reduce customers’ time waiting for answers.
“This is our way of applying technology to the high-touch side of the bank’s business,” says Ross.
Read more: Retail Bankers Are Adopting AI for All the Wrong Reasons
How GenAI Could Influence Erica 2.0
Ellis says the bank’s tech teams are actively planning the next couple of years of improvements to the Erica consumer app, though there is no set timetable yet.
“That’s partly because we’re still scaling the capabilities with our customer-facing associates and making sure we get good feedback there,” says Ellis.
He says this is an example of the bank’s “go slow to go fast” philosophy. Ellis explains that trust is a “big deal for us with our clients” and the bank wants to be sure GenAI capabilities work well before putting them into applications that directly touch the customer without a human in between.
Ellis couldn’t get into specific features anticipated for Erica 2.0. However, he says, “you can imagine where we will go by looking at some of the things you’re able to do on AI platforms today.”
Key strategic insight: A key hope for an Erica 2.0 augmented by GenAI is better use of customer data.
“There’s more opportunity in terms of better understanding of our clients’ lives, so we can make really good recommendations,” says Ellis. “With all the new technologies, there’s going to be more and more opportunities to do that. It’s going to be a continuation of the journey we already started.”
Read next: Who Will Protect Banking Consumers’ Rights in the Age of AI?
