Meta, Klarna scale back AI workforce cuts after productivity fails
Meta halted further company-wide layoffs and Klarna resumed hiring human customer-support staff after internal evidence suggested their AI systems had failed to deliver the productivity and service-quality gains executives had promised, according to reporting on internal restructuring plans and hiring reversals across 2025 and 2026.
Meta's AI-Native Pivot Stalls
Meta CEO Mark Zuckerberg led an internal restructuring plan centered on replacing staff with AI agents, following an initial reduction of approximately 8,000 employees—roughly 10% of Meta's workforce—in May 2026 Agentry. Internal data showed that despite a 220% increase in AI-generated code output, the productivity gains did not translate into comparable user-facing feature releases or measurable efficiency improvements Smarter Articles. The company subsequently canceled a planned second wave of company-wide cuts after the initial May reductions.
The gap between code volume and actual product delivery suggests that agent-generated work required substantial human review and refinement to reach shipping quality—a finding that contradicts the cost-reduction thesis underpinning Zuckerberg's original AI-native strategy.
Klarna Reverses Course on AI Customer Service
Klarna CEO Sebastian Siemiatkowski told Bloomberg in May 2025 that the fintech company had resumed hiring human customer-support staff so customers could reach a person Stewie. The rehiring followed an earlier period in which Klarna had reduced human support headcount in favor of AI-driven customer interactions. The move signals that Klarna's AI systems had not met customer-satisfaction benchmarks or handled complex service scenarios at the quality level required to sustain an all-AI support model.
Implications for the Agent Economy
Both reversals reflect a pattern emerging in October 2026: early-stage agent deployments have consistently overpromised on automation and cost savings. The failures were not technical incompetence but rather a mismatch between agent capabilities in controlled environments and real-world operational demands—handling edge cases, maintaining brand voice, and managing customer emotion in support contexts.
Neither Meta nor Klarna has published formal post-mortems or released performance metrics. However, the hiring reversals constitute concrete market feedback: companies betting on rapid workforce replacement via AI agents are discovering that human oversight, judgment, and presence remain irreducible costs of operation at scale. The reductions demonstrate that the agent economy's first wave of mass adoption is correcting toward hybrid human-agent models rather than full displacement.