Customer service automation has become one of enterprise tech’s most AI-intensive workloads. The shift goes beyond chatbots to multimodal interactions, proprietary training data and infrastructure handling hundreds of billions of application programming interface requests annually.

Cloud-based software company Zendesk Inc.’s approach is built on a multifaceted relationship with Amazon Web Services Inc., spanning infrastructure, voice capabilities and large language models, according to Tom Eggemeier (pictured), chief executive officer of Zendesk, which provides customer service and customer relationship management platforms.

“We have like a three-legged stool relationship with AWS,” Eggemeier said. “Number one, we are a customer of theirs, and we have all of our infrastructure on AWS. We’ve had it for 10-plus years. The second thing is [that] we are leveraging Amazon Connect, the voice platform, to integrate with our suite of products. The third thing is that we’re working with Amazon … Bedrock overall, where they’re getting all the large language models because AI is becoming an increasingly important part of our business.”

Eggemeier spoke with John Furrier and Paul Nashawaty at AWS re:Invent, during an exclusive broadcast on theCUBE, SiliconANGLE Media’s livestreaming studio. They discussed how Zendesk is leveraging AWS infrastructure, Amazon Connect and AI models to drive customer service automation.

Nova models fuel customer service automation

Zendesk’s work with Amazon’s Nova models reflects a broader re:Invent theme around open, flexible architecture. The approach allows enterprises to pair AI models with their own customer data, according to Eggemeier.

“We really love the Nova models, the fact that you can usually voice imaging and text because we think people experience with companies on a multimodal basis,” he said. “We have 18 billion interaction data points where customers rated thumbs up or thumbs down,” he said. “We really see the promise with Nova to be able to use … that data to make sure that we’re training the models and even in a better way.”

The AI push has transformed Zendesk’s business, with AI products generating $200 million in revenue over the past two years, according to Eggemeier. The company now has 20,000 customers using its AI capabilities.

“Five billion automated resolutions — that is an AI agent actually solving the problem,” Eggemeier added. “We don’t get paid on [a] consumption basis. We only get paid if the AI agent solves the problem.”

Here’s the complete video interview, part of SiliconANGLE’s and theCUBE’s coverage of AWS re:Invent:

Photo: SiliconANGLE