Cisco’s new AI agents and assistants aim to ease customer service headaches

Even in today’s modern age, call center customer service continues to be a nightmare. Wait times can be long, customers get frustrated when caught in loops with limited and robotic chatbots, human agents are overworked and don’t always have the visibility into information they need, calls are dropped, information is lost … and the list goes on.

But AI agents are growing ever more sophisticated, showing promise in nearly every area of the enterprise, including customer service.

Cisco is angling to be a leader in AI-powered call center support, and Wednesday at its WebexOne 2024 event, it announced new Webex AI agents and assistants that will work alongside humans to streamline processes and address common headaches and snag points.

“In this age of AI, there’s one belief that we hold close to our hearts, and that is the value of human interaction,” Anurag Dhingra, SVP and GM of Cisco collaboration, said in a pre-briefing. “Experiences matter, and in fact matter more in the age of AI.”

AI helping humans

Over the last 18 to 24 months, Cisco has been working to improve the support experience for both customers and employees, said Dhingra. He pointed to a survey the company did with 1,000 customer experience leaders in 12 industries and 10 global markets. More than half of respondents (60%) said self-service isn’t working, and 60% also reported that human agents are overworked. Further, customer experience leaders reported that 1 out of 3 of those agents lack the customer context needed to deliver the best possible customer experiences.

“A lot of this is rooted in fragmented and siloed investments,” he said (indeed, 88% of respondents to the Cisco survey reported technology siloes).

Webex AI Agent, available in early 2025, will bring together conversational intelligence with generative AI to help improve self-service. Dhingra pointed out that, while customers want self-service, 55% avoid it because they find it “rigid and unhelpful.”

Existing AI bots “tend to be scripted, robotic, quite limited in what they can do,” said Dhingra.

The AI Agent platform will include a new design tool, AI Agent Studio, which allows contact centers to build, customize, and deploy voice or digital agents in minutes, choosing the model of their choice, Cisco said. This could help improve customer interactions and resolve issues more quickly and easily.

Dhingra pointed out that well-functioning self-service options can result in a 39% improvement in customer satisfaction (CSAT) scores. Furthermore, predicted Jeetu Patel, Cisco’s EVP and chief product officer, in the near future, 90% to 95% of customer calls will be handled by automated agents. But, he said, “it won’t feel like you’re talking to a robot, it will feel like you’re talking to a human being.”

Human interaction still critical

However, Dhingra pointed out, “there is no substitute for humans; sometimes customers need to talk to humans.”

AI Assistant for Webex Contact Center (to be generally available in Q1 2025) will work alongside humans to quickly supply information. New tools will include:

Suggested responses that appear directly in the platform.

Context summaries that help facilitate handoffs from AI to human agents. Necessary background information is included to get the human agent up to speed and eliminate the need for customers to repeat themselves.

Dropped call summaries: All interactions are captured and documented so that agents and customers can pick up where they left off when a call is resumed.

Automatic CSAT scoring based on data and transcripts of customer interactions.

In addition, to support overworked humans, Cisco plans to release an agent wellness platform that will schedule automatic breaks and shift which channels agents support to increase or decrease capacity based on need.

“It’s around helping humans,” said Dhingra.

Able to pivot, understand limitations

Prior to launch, Cisco is performing an internal pilot of Webex AI Agent in its human resources department. In a demo video shared by Dhingra, a worker interacted on her phone with an AI agent and asked the agent to book time off for her in Workday. The agent asked how many days off she was planning to take, and when, then booked it.

The worker then asked the AI agent to send her a new laptop, to which it replied: “I’m sorry but this is not something I’ve been trained on. I’m still learning as an AI agent, thanks for your understanding. If you need a new laptop, please contact your IT support team.”

This ability of AI to understand its limitations is critical, said Dhingra. “There are guardrails in place to stop it from going off rails.”

In the demo, the employee also interrupted the agent a few times, asking one question, then quickly switching to another. After a quick pause, the agent was able to switch gears. “[The employee] pivoted, asked about one thing, before an answer was given, she changed gears,” said Dhingra. “The AI agent was able to deal with that quite effectively.”

Further, he pointed out, the voice was “a lot more natural sounding, not as robotic as it has been in the past.”

Responsible AI ‘table stakes’

Responsible AI is top of mind for nearly every enterprise leader today, and Dhingra explained that Cisco has developed a set of principles it uses to review its AI systems, looking at transparency, fairness, accountability, reliability, security, and privacy.

The goal is to provide customers with information on how the models work and how they were trained, to “describe how the AI works under the covers,” he said. “We break down our responsible AI by use cases.”

Customers can also provide feedback that Cisco treats “just like security incidents,” he said. If a customer flags an issue, there is what he called an “urgent process in place” that will look into it.

“I do think this is table stakes for everyone right now,” said Dhingra.