Cisco Systems is restructuring its $15 billion services division around agentic AI and workflow automation to eliminate human availability bottlenecks, CIO.com reported.
Every workflow constrained by human availability is a workflow for redesign, according to Liz Centoni, EVP and chief customer experience officer.
The multinational tech conglomerate is pushing an AI-native operational overhaul across its 20,000-employee services organization, which handles professional services, customer success, technical support, and software renewals. This internal and external transformation builds on Centoni’s leadership during the 2024 acquisition of Splunk for $28 billion, CIO.com reported.
The tool successfully accelerated the process of finding the right engineer, but it inadvertently exposed five internal handoffs to the customer, creating more frustration.
“We just bolted AI onto the wrong thing and said, ‘automation,’” Centoni told CIO.com.
Cisco realized that simply automating legacy workflows designed decades ago was insufficient. Instead, the team evaluated its 40-plus years of customer engagements to build a centralized knowledge base and structured playbooks. Rather than treating every customer engagement as a bespoke model requiring multiple expert engineers, deployable expert agents now democratize historical problem-solving data.
“You go into a new deployment with an expert agent that brings in all the prior experts and democratizes the knowledge base,” Centoni told CIO.com.
Traditional support metrics focused heavily on compressing root cause analysis times from hours down to minutes. The integration of agentic AI shifts that operational mandate from reactive speed to predictive prevention.
Agents now continuously monitor customer environments for telemetry data signaling emerging failures, applying historical case clusters to recommend next best actions before outages occur. Centoni stated an explicit 12-month goal to sit down with customers and show that out of ten potential outages, eight were prevented proactively.
Rising enterprise scrutiny over the return on investment of AI token consumption has forced a more rigorous approach to architecture selection. Cisco relies on a hybrid stack combining traditional predictive AI models, rules-based automation, and newer large language models depending on the specific operational constraint.
Data sovereignty and on-premises infrastructure requirements heavily influence these choices. For the internal AI assistant handling software renewals—which coordinates a renewals agent, sentiment agent, risk agent, and install base agent—the company selected Mistral.
“If you look at our AI assistant for renewals, which brings together a renewals agent, a sentiment agent, a risk agent, and an install base agent, it’s actually based on Mistral because they were one of the few who could give us an on-prem model that had the full capabilities,” Centoni told CIO.com.
Enterprise AI deployments often suffer from compliance theater, where users check adoption boxes on software dashboards while continuing legacy manual processes. Centoni measures success not by initial adoption rates, but by structural changes in operational capacity.
The internal support agent integrates directly into the initial case-opening workflow, automatically pulling relevant customer history and environment context. This eliminates the first 30 minutes of manual information gathering.
“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 told CIO.com.
