Amazon Connect VP on combining AI with human care in customer service
“You know, I never thought my life’s work would be working on contact centers,” Pasquale DeMaio said, “but it turns out to be one of the most interesting things I could have ever worked on.”
The vice-president of Amazon Connect at AWS was reflecting on a career journey that has not only focused on transforming the way organizations approach customer service, but helping them determine the most balanced approach to using artificial intelligence.
DeMaio has been overseeing the growth of Amazon Connect, the company’s cloud-based contact center platform, for the past eight years and as a result as seen it grow to thousands of users. This includes brands such as State Farm, Traeger Grills and Air Canada, which have deployed the technology to improve resolution times, customer satisfaction and more.
Earlier this year, the company announced a new generation of Amazon Connect whereby AWS embedded first-party AI it said would be activated in just a few clicks. The AI features span everything from customer self-service and agent assistance to conversational analytics.
That said, DeMaio said many Amazon customers are taking a measured approach to AI, with a heavy emphasis on keeping humans in the loop to avoid unexpected problems.
“You can have some pretty up and down experiences with hallucinations and stuff like that,” DeMaio told 360 Magazine. “I see very few large customers who are like, ‘We’re just gonna throw everything over to LLMs and hope for the best.”
Bringing greater visibility into customer service operations
One of the historical challenges in improving customer service is a fragmented or inadequate ability to see what’s going on, DeMaio said. He gave the example of a customer who calls up to cancel their subscription, only to have an agent badger them into keeping it. The same customer might simply call back the next day and cancel with a different agent. Yet only some of those calls might get recorded.
“The whole point of us bringing our all inclusive AI (into Amazon Connect) is you can actually measure every single interaction, every touch point, and then manage the impact with AI, whether that be helping the customer do self-service or help the agent by giving them more information,” he said.
Matching and merging customer profiles and centralizing the data has been among the biggest benefits AWS has gained through generative AI, DeMaio said. AWS has also been using LLMs to improve brands’ knowledge bases within their customer service operations. They can get a better sense of what’s being used, what’s not, and what needs to be updated or corrected.
The ‘better together’ scenario for humans and AI
“Our approach is very much focused on a better on a ‘better together’ scenario where humans are getting better and AI’s getting better together,” he said, adding that many organizations, such as a large U.S. bank he and his team work with, are beginning to recognize that automation needs to be aimed at boosting retention rather than simply reducing work within contact centers.
“They were saying, ‘We have no goal to lower the number of phone calls we get. We just want them to be a lot better,’” he said.
The rise of AI within customer service may call for a different outlook on the metrics that matter for CX leader, DeMaio said. He gave the analogy of a pair of running shoes: a consumer might have bought them with the intention to work out but then failed to do so. Part of an effective customer service strategy might be about helping them get more use out of a product they have.
“CSAT may have some interest, but I really want to look at how do these customers behave when we treat them fundamentally differently than we have historically,” he said. “Then, how do I optimize that with both the automated experiences and the human ones?”
