At Transform 2019 today, Ryan Lester, senior director of customer experience technologies at LogMeIn, said companies should not just think about AI from a bot perspective, but rather from multiple perspectives. And while many companies have begun exploring AI applications for customer service, he sees the current environment as ripe for advancement.

“It’s a unique opportunity to not think about AI as a pointed solution, not just chatbot, not just a search tool, not just a better way to understand the customer journey, but rather use it throughout the customer journey,” he said.

When it comes to implementing AI, it’s helpful to think of the virtual agents for customer support as slotting into two different use cases, according to Jonathan Rosenberg, CTO and head of AI at Five9. The first is a business segmentation, which is figuring out why the customer called, he said. The second is to provide a self-service.

Some customer requests are low-hanging fruit, meaning a bot can be easily deployed. For example, a leading use case at Hulu would be helping customers who are unable to log into a website, said Hulu director Matt Kravitz. A virtual agent could simply walk that customer through a login or account recovery procedure.

Other contexts might require a combined approach — like when a customer is about to cancel a subscription. A virtual assistant could determine why the customer is calling, but a human customer service agent would be better able to persuade that individual to stay.

“So you’re still using the system to segment, but you’re using a human being,” Kravitz explained.

Educational resources company Chegg, which offers services like online tutorials and textbooks, might also deploy an AI-powered business segmentation to ensure its services are relevant for students. To do that, it must first determine what the user needs, and then evaluate whether its immediate offerings suffice.

“Based on what the intent was, maybe we want to … have [users] call in to get more complex service. Or maybe we want to send them to an article online. Or maybe we want to have them chat with someone,” LogMeIn’s Lester explained. Traditional keyword searches might not work in a Chegg scenario because keywords are often contextual and not personalized to the student’s actual experience.

With an AI-powered solution, Chegg can route the user to the appropriate resource. A bot might be able to handle inquiries about textbooks, for example. But when it comes to signing up for a new service — where there may be additional questions — connecting students to a human over the phone might be most effective.

“But by having this all be powered by AI, we can be more selective in the results we’re given, the context sales results, and how we want to service that customer,” Lester said.