Cisco’s AI Pivot: Why Efficiency Isn't Enough

“It’s painful,” says Liz Centoni, describing the integration of artificial intelligence at Cisco as “surgery without the drugs.” As the chief customer experience officer, Centoni is overseeing a massive transition of her 20,000-employee division, moving beyond simple automation to fundamentally redesign how the tech giant services its global client base.

Cisco initially attempted to bolt generative AI onto existing workflows, but the strategy backfired. Early experiments in customer support—using AI to summarize case handoffs—merely accelerated inefficient processes, annoying customers rather than resolving their technical issues. Centoni realized that speeding up a broken workflow is not progress. The real challenge was ensuring that support cases reached the correct engineer on the first attempt.

By shifting focus to intelligent routing, the company now directs nearly 88% of its 1.5 million annual support cases to the right expert immediately. This metric represents a pivot in how Cisco measures success: prioritizing fewer handoffs over faster ones. Centoni advocates for applying AI to repeatable tasks where models can achieve 90% accuracy or higher.

To manage this transition, the company launched Cisco IQ, a digital interface designed to act as a single source of truth. The platform aims to identify preventable outages and free staff from data interpretation tasks. For Centoni, the mandate for any new AI initiative is clear: it must either grow revenue, expand margins, or deepen customer trust. If a project does not actively replace outdated work or contribute to future innovation, it does not make the cut.