AI Undergoes a Contact Center Expansion

Article Featured Image

Contact centers have been looking at artificial intelligence the wrong way for years, if not decades. They’ve long believed that if they just moved customer service functions to automation, they could solve their contact center staffing woes, save costs, and improve customer outcomes. They’ve long expected technology to provide all of the answers that customers might want while at the same time reducing their need for expensive contact center agents.

Turns out, though, that automation alone can only handle a small set of simple tasks, like changing passwords, checking balances, or tracking shipments, and there are enough more complicated customer issues that still require a full complement of contact center agents.

Companies can get the best of both worlds by following the latest trend—blending all three forms of AI—attended, assisted, and fully automated—in a way that ultimately streamlines workflows and positions automation as a real-time copilot for agents.

Powered by customer intelligence that is collected and analyzed in real time, using AI in this manner provides the resources contact centers need to eliminate many of the challenges both customers and agents face today.


Artificial intelligence has become a buzzworthy (at times overhyped) topic in just the past few years, but the fundamental principles of AI in the contact center go back several decades.

“Everyone is talking about AI right now, but it’s been around for a while in different flavors,” says Heather Richards, vice president of go-to-market strategy for digital-first engagement at Verint. Even as far back as the 1970s, the basics of AI were being developed, she says.

John Willcutts, general manager of the digital solutions group at NICE, agrees, noting that his company has been collecting customer information from interactions for 40 years. “We’ve been using that data to drive value for our customers—why did people call? What was the resolution? How do you train agents to perform tasks better? The data collected and analyzed is the essence of modern AI models,” he says.

Over the decades that have followed, very basic versions of AI became part of the contact center technology stack in other forms, like call routing, natural language processing, analytics, workforce management, conversational AI, and more.

And just within the past year, ChatGPT has ushered in an entirely new AI revolution, with generative AI upending everything across customer service, marketing, and sales functions. Within the contact center, though, its use is still limited, most experts agree.


Now contact centers have at their disposal many AI-based capabilities that allow them to provide fully automated self-service as well as assisted and attended automation. Industry experts recommend that, rather than going all in on one of these, contact centers look to a combination of attended automation, assisted automation, and full automation.

Attended AI interactions, which have humans in the loop, rely on AI to suggest insights and answers, but the agent makes the final decision about what to provide to the customer. Assisted AI interactions are similar; though AI is more involved, the human agent still has final control. Fully automated, as the term suggests, means that human agents aren’t involved in interactions at all.

Attended automation helps agents complete tasks in the course of their duties, like pulling up customer information during calls or chats or automatically filling in customer information. Unattended automation works behind the scenes to do things like processing information or importing data, explains Richard Atherton, senior product manager for CRM at BT.

Modern contact centers benefit the most when all three of these AI technologies work in unison.

“A lot of these things work better if they are all on the same AI platforms so that if you know what the customer wants, you don’t have to cross different platforms,” says Brett Weigl, senior vice president and general manager of digital, AI, and journey analytics at Genesys.

Behind the scenes, the goal of any AI in the contact center is to guide all interactions down the optimal path. It isn’t—or at least shouldn’t be—to eliminate humans in the contact center. Instead, AI should make contact center interactions as easy as possible for everyone—the customers, agents and leadership teams, says Fortuné Alexander, senior director of product marketing for customer service and sales automation at Pegasystems.

“Our vision is autonomous service with the ultimate outcome and making service effortless for everyone involved in the process, the customer and company alike,” Alexander explains. “Pega has a lot of technology that enables a high level of AI and automation to take you down this autonomous continuum.”


Even though Verint, NICE, Pegasystems, Genesys, and other vendors have embedded a high degree of AI into their modern contact center solutions, there are still plenty of contact centers using very old systems with technology that hasn’t been updated for decades, according to Alexander. “There is too much volume hitting contact centers because companies aren’t sophisticated enough with their tech stacks to implement self-service.”

Even those with self-service capabilities still fail when it comes to full self-service, he adds. “For example, if I go online and book a family vacation, then later I want to add my daughter’s 12-year-old friend, I have to buy a separate ticket using a separate record locator. I can’t just link them together and check her in because they think she’s a minor traveling alone because it’s booked on a different ticket. I have to call into an airline contact center, then spend an hour of my time getting that done.”

It’s particularly frustrating, Alexander notes, because the technology exists to avoid these kinds of complications. Rather than going through a live agent, why not enable the customer to use the same technology that the agent would use to accomplish the same end result?

Another big part of the problem is that companies have often designed their processes so that incoming interactions go the fully automated route first, which requires the customer to jump through several very frustrating hoops before the interaction eventually gets routed to a human for assistance. While this might deflect some of the mundane, routine, easily answered queries, such as “What is my balance” or “How late is the Main Street store open today,” such designs can have drawbacks when a customer truly needs to speak to a human being.

Weigl says Genesys has built its AI-based solutions with “escape valves” in the customer journey so that an interaction can easily transition from fully automated to agent-attended or -assisted, if necessary, without forcing the customer to go through too many hoops.

And then, responsible use of AI in contact centers, as in other business departments, is a concern, according to Jim Kaskade, CEO of Conversica.

A recent Conversica survey found that 73 percent of all organizations and 86 percent of those that have adopted AI agree on the importance of having clearly established guidelines for the responsible use of AI. However, only 6 percent of companies have policies in place for the responsible use of AI. And among organizations planning to adopt AI in the next 12 months, it’s even lower: Only 5 percent have policies.

Kaskade points out that while AI-powered chatbots fill the need for quick customer service as interaction volume grows, for some it has backfired, often because the AI is trained on outdated information.

AI models need to be trained on accurate, current data that comes from trusted sources; otherwise the AI will provide the customer or the agent (depending on the type of interaction) with inaccurate information, Kaskade maintains.

Another shortcoming with many current AI deployments is that companies don’t offer the right mix of automation, or at the right time and in the right context.

“You have to use the right AI (assisted, attended, or fully automated) for the correct use case,” Richards says. “If there is something that requires human oversight before a response, make sure there is a human in the loop at the appropriate place. That’s how you make sure that generative AI isn’t creating or hallucinating things on behalf of your business.”

Genesys, Weigl says, has taken on a different approach than some other vendors by layering in journey analytics. “We firmly believe that as you differentiate, you innovate AI and digital and layer on journey analytics to track where the data is going,” he says.

Following this strategy enables companies to find when and where customers abandon fully automated interactions, indicating that a change is needed to the fully automated process or that the interaction might be better suited for attended or assisted AI, at the specific point in the interaction where difficulty or dissatisfaction occurs, Weigl explains.

That’s why most AI solutions on the market today include real-time alerts to inform contact center agents and managers if an interaction has been handled to the customers’ satisfaction, Richards says. “For some use cases, fully automated self-service isn’t appropriate. Organizations are becoming more savvy in terms of how they choose to deploy self-service.”

AI can even be used for sentiment analysis and intent recognition to identify interactions that are best suited for self-service, assisted, or attended automation and execute the transfers.

To ensure that AI is delivering the expected results and benefits for each type of interaction, companies need to make sure they’re using the correct datasets, Richards says.


AI can help agents track down customer, product, or order/shipment information; automate repetitive, low-value tasks; and guide agents through workflows, processes, and next-best-action recommendations. This, in turn, allows them to spend more quality time with customers who truly need it. Research shows that this leads to happier, more productive agents, who are then more likely to stay in their current jobs.

But no matter what, inevitably some contact center agents are going to leave, and contact centers often struggle to train agents and bring them up to speed quickly enough. AI can help here, too, decreasing the time needed to train agents by guiding them when they need help and automating functions that previously required weeks of training.

Additionally, agents who have access to an AI-powered copilot need less time to learn to do their jobs well and can devote more of their time and attention to providing great customer service and less to toggling between systems or recalling procedures.

Ultimately, happier customers and agents translate to better metrics. AI delivers immediate ROI by improving the real drivers of contact center performance, like productivity and call quality, and impacts other metrics, like first-contact resolution, talk time, average wait time, and cost to serve.


Most industry experts expect very rapid growth of AI in all types of contact center interactions.

A study by Deloitte found that 53 percent of organizations are already using AI for customer service and 72 percent plan to increase their AI investments.

And research firm Gartner expects investments in contact center and conversational AI to dip in the short term “as business volatility creates a lengthening of decision cycles,” but sees steady growth in the long term, says Megan Marek Fernandez, a Gartner director analyst.

But even among the current economic turmoil, worldwide spending for contact center AI and virtual assistants is projected to total $18.6 billion in 2023, an increase of 16.2 percent from 2022, according to Gartner.

“Longer-term, generative AI and growing maturity of conversational AI will accelerate contact center platform replacement as customer experience leaders look to simultaneously improve the efficiency of customer service operations and the overall customer experience,” Marek Fernandez says.

However, Weigl says the Gartner outlook is overly optimistic and expects the growth to be much slower and more steady over the next few years.

And Richards expects companies to become much better at deciding how to best deploy the right AI for attended, assisted, and fully automated interactions. 

Phillip Britt is a freelance writer based in the Chicago area. He can be reached at spenterprises1@comcast.net.

CRM Covers
for qualified subscribers
Subscribe Now Current Issue Past Issues

Related Articles

Contact Centers See a New Focus on Onshoring

Foreign contact centers, once the darlings of the industry, have lost some of their appeal.

How to Build the Perfect Bot

New developer tools are steering automation creation.

Buyer's Guide Companies Mentioned