Klarna CEO Admits Overdoing AI, Rehires Humans to Restore Service Quality

Klarna's AI course correction: why the fintech is hiring humans again

Klarna cut deep on costs with AI. It worked-until it didn't. CEO Sebastian Siemiatkowski now says, "We probably over indexed," and the company is hiring again to restore service quality and product velocity.

After a strong US IPO-shares up 30% at debut, valuing the firm at $19.65 billion-the signal to the market is clear: growth and customer experience beat raw efficiency.

What changed

Klarna eliminated thousands of roles, dropped vendors like Salesforce, and leaned into AI assistants reported to be doing the work of 700 employees. The company even used an AI avatar of the CEO for earnings and an interactive avatar for hotline support.

The savings were real, but the trade-offs showed up in product and service quality. Over the last six months, leadership has been course correcting-rebalancing automation with human expertise and opening new roles.

Why this matters for executives

Cost cuts scale linearly. Customer trust does not. Over-optimization for efficiency can degrade the experience, churn high-value accounts, and stall product learning loops.

Public markets reward credible growth and resilient service. Post-IPO, the narrative must shift from "cheaper" to "better and faster." AI remains a lever-just not the whole operating model.

Strategy signals to watch

• Hiring mix: product, risk, merchant ops, and customer success-where judgment and context create advantage.
• Quality metrics: NPS/CSAT, first-contact resolution, AHT trend, retention, and merchant activation velocity.
• AI governance: clear escalation paths, human-in-the-loop at breakpoints, and error budgets for automated systems.
• Vendor posture: selective reintegration where third-party capability outperforms in-house AI stacks.

Executive takeaways

• Set dual North Stars: unit-cost efficiency and experience quality. Prioritize both in planning and reviews.
• Automate by stage-gate: pilot, compare against human baselines, then scale-don't rip-and-replace overnight.
• Protect critical moments: fraud checks, refunds, chargebacks, escalations require human judgment.
• Measure end-to-end value: savings are trivial if LTV, conversion, or merchant retention slide.
• Tell the right story to investors: efficiency as a means; growth, product cadence, and service as the ends.

90-day action plan

• Audit where AI replaced headcount; map to shifts in NPS, churn, and complaint categories. Restore coverage where thresholds were breached.
• Define stop-loss triggers: if quality dips below target, route to humans automatically.
• Rebalance teams: rehire or retrain in CX, merchant success, and risk ops; create "AI + service" pods.
• Instrument feedback loops: tag AI errors, track fix time, and feed learnings into product backlog weekly.
• Upskill leadership on AI operations and governance. If you're aligning roles to new workflows, review AI learning paths by job function at Complete AI Training.

Context

Klarna's shift follows aggressive automation moves, including an AI assistant reported to replace work from 700 agents. For background on that rollout, see Reuters coverage here.

The pivot doesn't reject AI. It reframes it: automate the predictable, staff the consequential, and compete on experience and speed-especially after you go public.