Wayfair and Zendesk CTOs on how gen AI will transform CX

Customers demand good digital services — in fact, they consistently judge companies based on the experiences they’ve had with them in the digital realm. 

So now, as enterprises increasingly adopt generative AI tools, noted Adrian McDermott, CTO of Zendesk, they must do so with an eye on enhancing customer experience (CX). 

McDermott spoke at length on next-gen CX at this year’s VB Transform alongside Wayfair CTO Fiona Tan in a panel moderated by Danny Tomsett, CEO of digital human platform UneeQ. 

Ultimately, McDermott forecasted, expect “huge change in the dynamic of the CX industry.” 

AI offering ‘personalization at a whole new level’

Tan noted that home decor, particularly, can be an emotional, personal category. Often, it can be difficult for customers to explain or describe what they’re looking for. 

To help consumers identify the styles they like and digitally redecorate their living spaces, Wayfair recently introduced a gen-AI powered tool Decorify. Users just upload photos of their rooms and try out different virtual decors, Tan explained. 

“You can see your living room in different styles,” she said. “It could be styles that you’ve never really understood or could articulate, and that you can now visualize. Maybe you find out you don’t like glam, perhaps you like mid-century modern.”

From there, Wayfair applies its product understanding to suggest real-life items in their catalog. Tan noted that the app has already generated 150,000 images, and that the people who engage with it convert much better. 

“It’s a very high-converting capability,” she said, and the company is working to incorporate it into their main website. 

“We can now generate that much more personalized content on the fly,” she noted. 

Tomsett agreed, marveling that gen AI offers “personalization at a whole new level,” and is a “super interesting new way to think about customer experience.”

Empathy will be ever more critical in the age of AI

Returns and exchanges are undoubtedly a big part of retail, Tan noted, and Wayfair has been able to employ human-in-the-loop AI in that process. For instance, when a customer submits a photo of a damaged product, AI can analyze it to help determine the type of return the customer needs. 

Oftentimes, returns can be complicated, she pointed out; people want human empathy, as in “You got this thing you’ve been waiting on forever and it’s got a huge scratch in it.” At least for now, “for critical applications, it still behooves us to have a human in the loop.”

However, it’s important, she emphasized, that enterprises allow consumers choice when interacting, offering both self-serve and human interaction. “We very much still want to make sure the human gets to choose,” said Tan. 

Tomsett agreed, noting that “not every consumer is the same and so we shouldn’t treat them the same.” 

He explained that his company works with many large enterprise customers, including in telecom. “Telco services can be really straightforward for some people, and really intimidating for others,” he pointed out. 

Surprisingly, more than 40% of consumers feel intimidated when interacting with telecom customer service. “It’s a real emotion that’s not being met,” he said. 

To help address such trepidations, UneeQ incorporated their digital human Sophie into one telecom customer’s website; she interacted with users, validating that the process can be intimidating and calmly asking them to explain the issue they were having in their preferred language. 

People often feel a risk of judgment, Tomsett noted, and sometimes feel more comfortable talking with a digital human. In fact, the telecom customers that incorporated Sophie saw 5X higher conversion. 

“Empathy and the ability to detect sentiment are going to become more important,” said McDermott. 

He noted that enterprises will be able to automate some human conversations in CX, but they should also be using AI to amplify human skills. 

“There’s always going to be a percentage of the population that is distressed, intimidated or just likes to press zero until they can speak to the operator,” he said. 

To this point, CX has been measured “like a machine; we look at the throughput, volume, speed.” With AI, it’s going to be more about looking at quality and sentiment, he said. 

“The skill package for customer service agents is going to change dramatically in the next few years,” said McDermott. He noted that it will be less about efficiency, and “a lot more about empathy and the ability to deal with complex problems.”

Ultimately, large language models (LLMs)  have incredible world knowledge and an “extraordinary ability to predict,” noted McDermott. Their key skill is the ability to reason, but “we’re only just beginning to trust applications that provide scaffolding, governance and leverage that reasoning. That’s going to be the next huge transition in customer experience.”

Still, in the end, he emphasized: “You can only move as fast as your consumer population wants to move.” Retailers and brands must ensure that they’re “serving people the way they want to be served.”