Customer service is undergoing a radical shift in the artificial intelligence era, especially when it comes to contact centers, where efficiency has always been a battle.

IBM Corp. is supporting businesses’ customer service divisions with agentic AI. The company has focused on improving the contextual and transactional data that an agent needs to do its job.

IBM’s Glenn Finch talks with theCUBE’s Dave Vellante about agentic AI’s impact on customer operations.

“The bottom line is if as you embed an agent into a business process, the model needs context,” said Glenn Finch (pictured), global managing partner of cognitive and analytics at IBM. “I have the ability to now enable models and agents with context in a very short period of time. Where last year we’d be looking at multiple months to deploy agents, we’re now seeing agents deployed in a couple of weeks. It’s a radically different paradigm.”

Finch spoke with theCUBE’s Dave Vellante at the “AI-Powered Business Operations: Strategies for End-to-End Transformation” event, during an exclusive broadcast on theCUBE, SiliconANGLE Media’s livestreaming studio. They discussed how IBM approaches contact centers’ need for efficiency and the evolution of agentic AI. (* Disclosure below.)

Adapting LLMs to contact centers

Contact centers are infamous for being slow and ineffective at answering customer queries. Finch believes that AI agents could take the pressure off customer service workers by ingesting research to answer common questions.

“For the last few decades, we’ve been doing machine learning, and machine learning is awesome with numbers,” Finch explained. “Then we went to NLP, natural language, and that’s what a lot of the previous things were built on. But generative AI takes it to a whole new level. When you think about it, its ability to speak, understand, understand inflection, things like that, it makes it ripe with all the conversational spots in customer contact.”

IBM has introduced InstructLab, which allows users to adapt large language models to their specific needs. This takes care of the contextual part of the AI data equation. The other half is transactional data, which can be trickier to process, according to Finch. IBM is currently experimenting with data virtualization techniques to adapt models and agents to transaction analysis.

“Up to a certain point, all the virtualization in the world works awesome until you have a hundred million customers, then it’s hard,” he said. “That’s when you start to see a little bit more hallucination from the model … but we are using more virtualization techniques, more proxy techniques, things like that to get the model both the context it needs and the data it needs to do its job.” 

Creating the generative user experience

With customer service, IBM has two priorities: efficiency and efficacy. Despite some bumps in the roads, clients have gradually embraced agentic AI and automation. Finch highlights how digital agents can stem the high turnover rate in contact centers, transforming a company’s customer operations.

“Clients are moving more and more towards being willing to let more things happen in a highly automated fashion,” he said. “A lot of clients are saying, ‘OK … we’re going to let, not a chatbot, but a digital virtual agent that can have a generative conversation with you, we’ll let them take calls.’ The strange part is what we’re finding is that the Net Promoter Score, the client perceived value of those channels is actually higher than the call center.”

Now IBM is taking agentic AI to the next level by not only answering customer queries, but creating a generative user experience, or gen UI. Engineers feed in the context of user transactions into an agent, which can then respond to a user with relevant details about the subject matter in addition to answering the user’s question.

“We have a client that has a voracious web and mobile experience,” Finch said, as an example. “But there are about 15 million times in a year that a client goes to the web and mobile and they don’t get what they want, so they pick up the phone. What we did is, we used gen AI to reverse engineer all those 15 million phone calls and build a generative user experience at the web and mobile.”

Here’s the complete video interview, part of SiliconANGLE’s and theCUBE’s coverage of the “AI-Powered Business Operations: Strategies for End-to-End Transformation” event:

Plus, don’t miss the event entire episode:

(* Disclosure: TheCUBE is a paid media partner for the “AI-Powered Business Operations: Strategies for End-to-End Transformation” event. Neither IBM Corp., the sponsor of theCUBE’s event coverage, nor other sponsors have editorial control over content on theCUBE or SiliconANGLE.)

Photo: SiliconANGLE