日本語
한국어
Tiếng Việt
简体中文
Bank of America is taking its artificial intelligence capabilities to the next level. The financial giant is rolling out generative AI (GenAI) enhancements to EricaAssist, the internal AI tool utilized by its contact center employees. This tactical move serves as a testing ground before the bank integrates GenAI into the customer-facing version of its virtual assistant, Erica.
Originally launched in 2018, the customer-facing virtual assistant Erica has grown exponentially, using open-source models trained on millions of client interactions. EricaAssist was introduced in 2020 to support staff. Now, by upgrading the internal tool first, Bank of America ensures its human agents can evaluate and guide the AI’s capabilities before they reach the public.
The Strategy: Keeping Humans in the Loop First
The decision to deploy GenAI internally before launching it to customers is highly intentional. According to bank executives, starting with contact center agents ensures there is always a “human in the loop” to review outputs, make final decisions, and provide feedback. This phased approach aligns with Bank of America’s “go slow to go fast” philosophy, prioritizing consumer trust and security over rapid, unchecked deployment.
Erica’s current reach is massive, highlighting the high stakes of any major update:
- The customer-facing assistant reached 24.6 million users in Q2, representing a 23% year-over-year increase.
- Erica handled over 200 million interactions in that quarter alone.
- Since its 2018 launch, the assistant has managed more than 3.6 billion customer interactions.
- Surprisingly, developers found that customers heavily prefer texting with Erica over using voice commands.
How GenAI is Transforming EricaAssist
The integration of GenAI is designed to remove tedious administrative tasks for the bank’s 18,000+ customer service representatives, allowing them to focus entirely on customer care. Operating as a persistent desktop widget, the upgraded EricaAssist automatically provides representatives with key details about a customer’s relationship and the nature of their inquiry as soon as a call connects.
Eliminating Manual Call Summaries
Previously, contact center agents had to manually write summaries of every customer conversation while simultaneously trying to resolve problems. Now, the GenAI engine drafts real-time, highly accurate summaries of the calls. Agents can quickly review, edit, and save these summaries without being distracted during live conversations.
Solving Complex Inquiries Instantly
Understanding why a credit card transaction was declined can be incredibly complex, with upwards of 1,000 potential reasons. Before the GenAI upgrade, agents had to navigate multiple databases to find the cause. With GenAI, EricaAssist pulls this information in seconds, allowing agents to resolve issues rapidly and utilize the remaining call time to address other client needs.
Looking Ahead to Erica 2.0
While Bank of America has not announced a specific release date, plans are underway for a consumer-facing “Erica 2.0.” The next-generation assistant will leverage GenAI to provide deeper personalization, a better understanding of customer financial habits, and highly tailored product recommendations. By gathering feedback from its call center employees today, BofA is building a more robust, reliable, and intelligent assistant for tomorrow.
Source: thefinancialbrand.com