Artificial intelligence news.
Straight from the original announcements.

Cisco’s Liz Centoni says users must help shape agentic AI

November 13, 2026, 14:00By the AI News Bites Editorial TeamSource: Cisco
Cisco’s Liz Centoni says users must help shape agentic AI
Image: Cisco

Cisco is combining agentic AI, traditional AI and automation to redesign its internal operations and customer services, according to Liz Centoni, the company’s executive vice president and chief customer experience officer. She says successful deployments must begin with the work itself and include the people who use the technology.

Key points
  • Cisco is redesigning services workflows with agents, traditional AI and automation.
  • Centoni says users and subject matter experts must help design workflows.
  • Cisco is using AI to detect possible customer issues before they become critical.
  • The company selects models and tools according to each use case.
  • Cisco measures workflow change and team capacity rather than AI adoption alone.

Redesign the workflow before adding AI

Centoni leads Cisco’s $15 billion services business, which has 20,000 employees across professional services, customer success, technical support and software and services renewals. She says the team initially made a mistake by building an agent to find the right engineer for customer problems. The tool accelerated a process that still involved five internal handoffs, making the experience more frustrating for customers.

“We just bolted AI onto the wrong thing and said, ‘automation,’” Centoni says. Her team instead examined why the handoffs existed and how the workflow could be redesigned around the desired result. The approach draws on 40 years of customer engagements, which Cisco has used to build a knowledge base and playbooks for AI agents and human experts.

Centoni argues that “Users are an essential part of the technology stack.” Subject matter experts and users should help outline workflows, she says, while teams must define where an agent can act autonomously and where a human should take over. Cisco’s goal is to use prior expertise to support customer deployments while keeping human experts available for discussions about improving the outcome.

From faster responses to preventing problems

Cisco is also using AI to move beyond reducing the time needed for root cause analysis. Centoni says agents can monitor customer environments for signals that an issue may be developing and recommend next steps before it becomes critical.

The company uses case histories and decades of knowledge about potential failures to identify recurring clusters. Centoni says her goal is that, in 12 months, Cisco could tell a customer how many outages might have occurred and how many were prevented.

She also says organizations should choose technology according to each use case. Cisco first asks whether automation can handle a task. Rules-based work can use automation, while tasks that combine multiple signals to make a prediction may require AI. Cisco also combines older predictive AI models, automation and newer large language models.

For its AI assistant for renewals, Cisco uses a Mistral model in part because it could provide an on-premises model with the required capabilities. The assistant brings together renewal, sentiment, risk and install base agents, with the model customized through a private instance and used alongside automation.

User feedback is a measure of whether AI works

Centoni rejects adoption as the main measure of AI success because users may simply report that they tried a tool while continuing to use older processes. Instead, Cisco examines whether workflows have changed and whether the renewals team has gained capacity since deploying its AI assistant.

She describes agents as continuous work in progress. “With an agent, it’s 60% of the way there in its first iteration in terms of usability, and the user is part of the tech stack that gives us feedback,” Centoni says. A feedback channel is therefore central to making an agent usable at scale.

Cisco’s internal support agent operates within the support workflow. When a customer opens a case, it retrieves relevant history and context for the support worker, then provides a next best action. For users, the approach promises less time spent gathering background information and more time helping customers, while for Cisco it establishes workflow change and user feedback as tests of whether its AI investment is delivering value.

Related news

Cisco Highlights Partner Opportunities Around Cisco IQ
Products and toolsNovember 13, 2026, 17:22

Cisco Highlights Partner Opportunities Around Cisco IQ

Cisco is highlighting how partners can build and deliver services around Cisco IQ, including a managed-service option now available through Cisco IQ Partner Support…

Cisco Reports Record $18.5 Billion First Quarter Revenue
Chips and infrastructureNovember 12, 2026, 22:05

Cisco Reports Record $18.5 Billion First Quarter Revenue

Cisco reported record first quarter revenue of $18.5 billion, up 24% year over year, while raising its fiscal 2027 revenue and non-GAAP earnings guidance. The company…

Business
BusinessNovember 12, 2026, 14:00

Cisco says AI infrastructure revenue will double this fiscal year

Cisco reported fiscal first-quarter earnings that topped estimates and said revenue from its AI infrastructure business will double this fiscal year. Shares rose in…