Guide to using advanced analytics and AI in business applications
A comprehensive collection of articles, videos and more, hand-picked by our editors
The hype train driving talk of artificial intelligence over the last couple of years is showing no signs of slowing down, but behind the unbridled enthusiasm, some businesses are finding that paths to real business value exist when AI's limitations are understood.
One area where businesses are finding the most value from AI today is in customer service. Chatbot applications are among the most mature areas of AI. But enterprises are finding that, while AI chatbots can provide value, they have to be deployed the right way.
For online test preparation company Magoosh Inc., that means giving machines license to recommend responses to simple customer service queries, while still maintaining a team of agents who handle more complicated issues.
Magoosh uses a customer service bot from DigitalGenius to handle incoming customer service inquiries. The system scans messages for their content and recommends prewritten responses that can be personalized or sent out as is. Ultimately, Magoosh plans to enable the system to send responses automatically, without requiring that a human agent review its suggestions.
AI chatbots don't match up on empathy
The bot approach works for most low-level customer issues, like asking for a password reset. But every once in a while, a customer has an issue that requires a human touch. Briana Cahn, student help manager at Magoosh, said that might be when a customer is feeling frustrated with an aspect of the service or is experiencing a previously unseen technical issue.
Magoosh has been using the DigitalGenius platform since January, and Cahn said the team plans to expand its functionality. But she sees ensuring that people are around to handle potentially emotionally sensitive issues as one key to the company's successful use of AI.
"It's been a slow progression," Cahn said. "It doesn't feel like we've reached this holy land. We've slowly integrated pieces of it. For us, I think it's a great way to blend the two -- using AI and the mastery of our support agents."
Bots still can't brand
For slightly different reasons, the customer support team at Dollar Shave Club keeps humans around despite using customer service AI chatbots. In their case, it mostly has to do with messaging.
The company implemented a chatbot called Answer Bot from Zendesk in April and is using it to handle simple customer service requests, like placing orders or answering questions about how a product works. For these queries, the bot automatically sends a link to a support page.
Brian Crumpley, manager of analytics and insights for the company's customer engagement team, said this has freed up time for some members of the customer support team to do more impactful work than answering simple emails. They now work on an in-house, magazine-style publication featuring customer feedback success stories. He said the stories help product managers understand how customers are using the products.
Support team members can also spend more time supporting Dollar Shave Club's brand. Crumpley said the company tries to give its brand a unique voice that incorporates pop culture and social media trends, which would be hard for an AI application to replicate in a natural way. Whenever a customer has a more complex problem or reaches out with general feedback, support agents can now take more time to respond in a way that reinforces the brand voice.
"What AI is doing for us is giving us a chance to have those conversations by automating other areas," Crumpley said. "I don't know how you'd teach a machine to do that. We want to introduce automation only when it makes sense."
Despite promise, AI apps aren't without controversy
Ready or not: Enterprise AI chatbots are on the way
Prebuilt algorithms lower the bar for chatbot development
AI-powered chatbots fuel in-app customer support