Agentic AI, at its best, delivers a significant experience improvement over earlier automated self-service options.

Brian Cantor, Managing Director of CMP (Customer Management Practice), noted during last week's CMP webinar that early AI forms "sounded a bit more human than static FAQ pages," but Agentic AI's ability to process and adapt to context makes the caller experience more natural.

Despite Agentic AI's promises, for businesses and customers alike, it oftenfalls short of expectations. Experts believe that to ensure success, businesses need to streamline data collection and storage processes; leaders also need to clarify where Agentic AI can replace human support and where it should serve as an assistive role to human service.

Cantor believes the key to success lies in approaching Agentic AI with the right mindset.

Cantor stated that the technology to create excellent experiences already exists, but the lack of a proper framework and vision is hindering leaders. "If you're not thinking about it the right way, it's a very negative signal for how you're going to care for your customers."

Data and People: The Core of Agentic AI

Entering the Agentic AI mindset is essentially an information problem. Businesses need to keep customer data collection and organization orderly, while leadership needs a clear understanding of contact center staffing.

Merlin Bise, CTO of AI solutions provider Inbenta, believes that for Agentic AI, every piece of customer data has value. Even small details can accumulate into a more personalized journey for the overall experience.

"Any information you can get is leverage," Bise said. "So any information the customer is willing to provide, you should collect—whether it's an address, email, or phone number. That information needs to be properly protected. But the more information, the more power."

Once businesses obtain data, teams need to store it in a manner that issecure and compliant with privacy regulations. The relevant data also needs to be properly structured so that AI can access it.

"Simply connecting OpenAI, Anthropic, or Google to a simple database within an organization and expecting magical results will never happen," Bise said. "If you one day want to achieve true Agentic AI but place reasoning models on top of poor data, its behavior will be entirely dependent on what you feed it."

Bise pointed out that successfully deployed Agentic AI willreplace some human support, but the technology has its limitations. Leaders need a clear understanding of contact center team members and their responsibilities to judge what Agentic AI can and cannot do in maintaining or improving brand customer service performance.

"There are always things you cannot do," Bise said. "That's why it's crucial to get buy-in from all the people who impact the business."