In 2024, Klarna CEO Sebastian Siemiatkowski told the world his AI chatbot was doing the work of 700 full-time customer service agents. The announcement went viral. It became the reference case for every boardroom conversation about AI replacing human workers.
The chatbot was genuinely impressive. It handled 2.3 million conversations per month across 35+ languages. It resolved routine cases — refunds, return status, simple payment questions — in about two minutes, compared to eleven minutes for human agents. Klarna froze hiring for support roles for almost a year.
Then reality arrived. Customers reported the bot often acted more like a gateway to human support than a full solution. Satisfaction scores slid as automation pushed too far. A chatbot can answer "where's my refund" all day. It cannot read the tone of a frustrated customer who's three months behind on a payment plan and needs a human who won't just recite policy.
By mid-2025, Klarna reversed course. Siemiatkowski told Bloomberg that cost had become too dominant a factor in how the support team was organized, and the result was lower quality. For the same CEO who'd spent a year telling investors AI would shrink his workforce, it was a striking admission.
The company began recruiting remote, gig-style support agents — targeting students, people in rural areas, and loyal Klarna users. By September 2025, Business Insider reported Klarna was reassigning engineers, marketers, and analysts into customer support through an internal "talent pool." Siemiatkowski said the company had reached an "epiphany: in a world of AI nothing will be as valuable as humans."
Klarna wasn't alone. A 2026 Careerminds survey of 600 HR professionals found that 2 in 3 companies that made AI-driven cuts were already rehiring, and 52% started within six months. Forrester projects 55% of executives who replaced employees with AI will regret it within 18 months.
The lesson isn't that AI failed. It's that AI without human judgment isn't customer support. The winning model pairs AI speed with human empathy, not one instead of the other.
As Siemiatkowski put it after the reversal: "We focused too much on efficiency and cost. The result was lower quality, and that's not sustainable." — a lesson many other companies are now learning the same expensive way.