Kevin Wu is the CEO of Leaping AI. Leaping AI automates call centers with voice AI agents for customer service and lead qualification.
In recent years, AI has been the focus of the call center industry. Its ability to automate responsibilities or even replace call center workers has been both a common prediction and also a common fear for leaders and workers in this space.
New vendors are popping up every single day, and existing call center providers are wondering how to catch up and what this new normal means for their existing business.
My company, Leaping AI, creates human-like voice AI agents for the automation of common use cases, such as customer service, appointment scheduling and lead qualification, for larger call centers and enterprises.
Over the past weeks, I have spoken extensively to other companies in the space, and whilst we’ve often agreed that voice AI agents are already able to automate simple requests, we’ve also agreed that automating complex workflows is still elusive, meaning humans are still necessary.
What Is A Voice AI Agent?
A voice AI agent can be seen as a digital worker for a call center. It is able to field calls and speak to customers with a human-like, but AI-generated, voice. The conversation is natural, given that modern voice AI agents are powered by large language models, the same type of technology underpinning chatbots like ChatGPT.
It is possible to connect a voice AI agent to company-specific knowledge either by uploading a knowledge base, via a PDF, for example, or by granting the AI agent access to internal data systems via an API.
To create the impression that customers are actually talking to a human, it is possible for modern voice AI agents to be interrupted. Some voice AI agents also have their own background music or sounds, such as busy office sounds.
As an example, a large web booking company we worked with recently introduced a voice AI agent that can help with booking-related queries over the phone during and outside of office hours.
The Current State Of The Industry
From what I’ve seen so far in the industry, most deployments of voice AI focus on a specific set of repetitive use cases, such as tier-one customer service, appointment scheduling and lead qualification. For instance, one of our clients leverages AI agents to automate their inbound and outbound appointment scheduling calls.
In some cases, voice AI agents can automate some repetitive customer service requests. For example, a global travel experience booking site we work with uses voice AI agents to automate certain booking-related queries, such as cancellations and rescheduling requests. When working with clients, I’ve found that 50% of repetitive booking-related queries can be automated with AI without any human involvement.
Nonetheless, the other 50% of calls need to be routed to a human. Sometimes a customer would like to explicitly talk to a human. Other times, the customer has a request that voice AI has not been trained to do. And still, other times the AI is just not effective in following instructions and completing multiple tasks reliably one after another.
Even still, most enterprises I’ve worked with that leverage voice AI agents only do so for the first 20% of conversations or to automate specific workflows. With that, they maintain their existing human workforce to handle more complicated flows or whenever it is necessary to route the call to a human.
The Way Forward
Even though there have been great advancements in AI technology over the last two years, I believe voice AI agents can, at best, only automate up to 50% of requests due to limitations in intelligence and customer acceptance.
Enterprises should carefully evaluate the abilities of current voice AI agent technologies and come up with a strategy for how these agents can improve the efficiency of parts of their call center. These systems should co-exist and thrive alongside your current human workforce.
When doing this, it helps to work with an experienced technology partner that can help you evaluate what flows are suitable for AI and what flows are better left to humans. Once you determine this, even as the first calls go live, a rebalancing of what AI can and should do should take place. This is especially important as technology continues to improve.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?