Here’s How Far We’ve Come with Voice AI in Customer Service

When it comes to customer service, even in-person interactions can be unpleasant. And doing business over the phone is usually markedly worse, especially if there is a bot involved.

There is one fintech fighting that stereotype, however. offers a virtual call center agent tailored to the financial services industry. And you won’t find the company referring to this virtual agent as a bot. Instead, uses terms such as “empathetic,” “smart,” and “human-friendly” to describe its virtual agent Voca.


Voca implements an AI that has been trained by listening to an organization’s recordings of successful agents. Voca not only imitates the representatives’ responses, it also uses a human-sounding cadence and adds pauses and filler words such as “um.” The use case in the video below depicts a collections scenario. Other possible applications for Voca include lead generation, customer qualification, appointment scheduling, cross-selling, and customer retention.

Voca’s collections agent in the video sounds remarkably human, especially with such a common name, Sarah. Sarah pauses in all the right places, has sympathetic intonations, and understands David, her client, even when he doesn’t use proper English.

All of this is part of’s secret sauce. The company’s virtual agent leverages information from the call such as speech rhythm, tone, and the speed of the conversation to identify the customer’s intent and emotion. As the call progresses, the virtual agent can even pick up on clues that indicate that what the customer is saying is different from what they actually mean.

What’s lacking

Because of common fraud tactics such as phishing, society has been trained to never offer personal information over an incoming phone call. Figuring out a way for the customer to authenticate themselves without compromising their identity is a major hurdle here. In fact, this is such an enigma that digital identity is one of the biggest topics in fintech, and one that will persist.

Maintaining human cadence is a second item that needs to be considered here. This isn’t obvious in the demo above, but if you watch the company’s demo at FinovateSpring last year (which won Best of Show), you may notice an awkward pause before each answer. For some, the moment of silence may be just long enough to wonder if the caller understood their answer. This could cause them to repeat themselves and result in the voice agent and the customer talking over each other in an awkward exchange.

Despite the challenges present in voice-powered customer service, has created a powerful tool. Voice has come a long way in reducing friction for not only financial services companies, but also their clients. Additionally, the new adaptations of voice have created a more human-like experience, which is something many consumers crave in today’s digital era.