Contact center AI faces its resolution test as metrics fall out of step
SiliconANGLE Valentina Villamil ● Covered by 2 sources
Contact center AI is moving past pilots, and the real test is whether it fixes problems or just answers faster. That’s why old metrics are starting to look wrong.
Based on reporting by SiliconANGLE, Valentina Villamil — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
AI in customer service is leaving the demo stage behind, and contact centers are where the bill comes due. They pack together huge call volumes, expensive labor and those awkward customer moments where a bot can either help or make everything worse. That makes them one of the clearest places to see whether AI actually pays off.
The pressure is already changing how vendors talk about their products. Cisco is pushing AI agents across both collaboration and customer workflows, while cloud contact center vendors are rolling out voice agents at a steady clip. Gartner says conversational AI will cut contact center labor costs by $80 billion this year. Big number, but the harder question is what gets saved besides time.
That’s where knowledge management steps in. Bob Laliberte of theCUBE Research said contact centers are attractive because failures are so visible and a bad AI interaction can raise customer effort, hurt trust and damage the brand. In other words, a shiny bot that punts the problem back to the customer is not innovation. It’s a more polite kind of failure.
The old scorecards are already showing their age. Zeus Kerravala of ZK Research said average handle time and first call resolution mattered in the old model, but they don’t tell the full story if the problem stays unsolved. That pushes brands toward outcome-based scoring, where the result matters more than the clock. It also means companies need better data, better integrations and a willingness to redesign workflows instead of bolting AI onto a broken process.
Five9 and others have been leaning on implementation playbooks and voice AI agents to speed up time to value. But Laliberte and Kerravala both pointed to the same deeper constraint: the handoff between virtual agents and humans, plus the knowledge base underneath it. The winners, Kerravala said, won’t be the ones that automate the most. They’ll be the ones that make outcomes more consistent and build trust with employees and customers along the way.
My take — AI-written commentary, not fact-checked reporting
The industry loves talking about autonomy, but contact centers are a reminder that the boring stuff still rules. If the knowledge base is messy, the agent handoff is clumsy and the metrics reward speed over resolution, AI just automates the same old bad habits. That’s not transformation; that’s repackaging chaos with a nicer interface.
Read more about this at: SiliconANGLE
