AI Agents Are Now Calling Call Centers, and Human Workers Are Overwhelmed by Robot "Customers"

A new phenomenon in customer service: call center workers now face not only humans but also AI agents calling on behalf of consumers. The practice is spreading as AI agents such as Meta's Muse and Instinct emerge, letting users send a bot to wait on hold, negotiate a bill, or cancel a subscription without the hassle.

Although AI call volumes are still small, they are growing fast. Outsourcing firms, banks, insurers, and retailers are starting to prepare for the wave. The question is not just ethics but also authentication: how do you verify that an AI agent truly represents a customer and is not a scammer? May a company ask whether the caller is an AI? And how do you handle calls that could flood phone lines?

An insurance claims agent in California said he receives several calls from the same AI agent every shift, representing accident law firms. The AI is often impatient and struggles with detailed questions such as whether an airbag deployed or whether the driver worked for Uber. "If it's still talking, we have to stay on the line until it disconnects," he said. Ironically, his company's empathy scoring system reprimanded him for not being accommodating enough toward the robot.

At a major credit card company, a retention agent said AI calls often aim to map the system. "I was once asked, 'If I wanted to reach a representative like you, how would I do that?'" he recalled. Company policy forbids asking whether a caller is an AI, but it is usually obvious from unnatural intonation and pauses. Only account holders can close a credit card, so AI cannot perform certain functions.

James Bednar, VP of Product & Innovation at TTEC, said AI calls from third-party service providers for bill negotiation and subscription cancellation are increasingly common. However, the AI often fails challenge questions such as the name of a first pet or favorite teacher, so the system flags it as a potential scam. "There's a lot of discussion among clients about how to handle those calls," Bednar said. He added that AI call volumes can be very large, pushing some brands to erect stricter authentication walls that end up hurting human customers.

Mike Clifton and Max Schwendner, co-CEOs of Alorica, envision the need for a digital identity layer for AI agents. "There's a missing layer for personalization, and it doesn't exist yet," Clifton said. Without such a system, companies face security and data risks, especially for sensitive transactions such as bank transfers. Meanwhile, Michelle Vaccaro of Harvard Business School argues that warm, dominant AI agents can negotiate effectively. "They can be quite fluent," she said.

For Indonesia, the phenomenon is an early warning. The domestic customer service industry, from banking to e-commerce, needs to prepare for AI calls that may arrive through local channels. Existing data protection and digital authentication regulations do not specifically govern AI agents. Without readiness, companies risk a flood of automated calls that drains resources, while consumers could lose quick access to human help.

Looking ahead, a question arises: will we see AI agents battling one another in negotiation while humans just watch? Or will companies choose to close phone lines to bots, at the cost of slowing service for everyone? Either way, the era of the "robot customer" has arrived, and the customer service world must adapt.