Cisco exec Liz Centoni admits adopting AI is ‘painful’; says: How do I tell the right engineer that …
Cisco executive Liz Centoni has acknowledged that adopting artificial intelligence (AI) across a large enterprise can be a difficult process. The executive of the US-based tech major described the transition as “surgery without the drugs” and admitted that “It’s painful.” Speaking to Business Insider, Centoni, Cisco’s chief customer experience officer, discussed the challenges of transforming the company’s roughly 20,000-employee customer experience division into what she called an AI-native services organisation. She said the company learned that simply adding AI to existing workflows did not necessarily solve underlying problems.One of the key lessons came from Cisco’s customer support operations. The company initially deployed generative AI to create summaries when support cases were transferred between engineers during shift changes or when different expertise was required. While the technology sped up handoffs, it failed to improve the customer experience. “The outcome was never the handoff,” she said. It was “how do I get to the right engineer the first time around?”
How Cisco is shifting focus from faster handoffs to intelligent routing
The experience led Cisco to rethink its approach to AI implementation. Instead of focusing on improving existing processes, the company redesigned its workflow to ensure support requests reached the appropriate expert from the outset.Centoni said Cisco adopted “intelligent routing” to match cases with the right engineer on the first attempt. According to the company, its customer experience division handles around 1.5 million support cases annually, and nearly 88% are now routed to the correct engineer the first time.A Cisco spokesperson said the company measures customer service performance by how many support requests require only one handoff or none at all.Centoni said the company discovered that AI delivers the most value in repeatable workflows that can be performed autonomously with more than 90% accuracy. She added that AI projects should be evaluated on business outcomes rather than solely on efficiency gains.Cisco recently introduced Cisco IQ, a digital interface for support and professional services. According to Centoni, the platform is intended to serve as a single source of information for customers, helping identify preventable outages, reducing time spent interpreting data, and addressing recurring support issues.For Centoni, successful AI adoption requires organisations to determine not only what work can be improved, but also what work can be eliminated. She said every AI initiative should demonstrate whether it can increase revenue, improve margins, strengthen customer trust, or help teams focus on future development efforts.
