
Sophie Cheng discusses Sinch's research on the challenges of maintaining reliable AI agents in production environments.
“If your AI agent failed today, would you know before your customers do?” asks Sophie Cheng, Chief Marketing Officer at Sinch. In this Telecom Reseller podcast, Doug Green speaks with Cheng about Sinch’s new research on what the company calls the “AI production paradox.” While many enterprises have moved beyond AI pilots, the bigger challenge is keeping AI agents reliable once they are live. Cheng says Sinch surveyed more than 2,500 global enterprise executives and found that 62% already have AI programs in production. But 74% also reported having to roll back a live AI agent because of operational reliability issues, including exposed data, hallucinations, tone-of-voice problems, auditability gaps and compliance concerns. The discussion focuses on why AI agents often perform well in controlled testing, but encounter problems when exposed to real customer interactions. Cheng says enterprises need stronger governance, clearer guardrails and communications infrastructure that can support secure, reliable, multichannel customer engagement. “More monitoring means more failures are detected,” Cheng says. “It does not mean more failures are created.” Cheng also explains why…
Host: Doug Green
Guest: Sophie Cheng
Organizations: Sinch
Explore listener stats, chart rankings, contacts and more on the Telecom Reseller / Technology Reseller News podcast page.