How Klarna’s AI did the work of 700 agents — then rehired humans.
The Klarna AI numbers.
Every figure below carries its named source and a grade: A = peer-reviewed/regulatory, B = top-tier press or primary company document, C = company self-reported (not independently audited).
A bold 2024 claim, a 2025 correction.
In February 2024, Klarna and OpenAI announced that Klarna’s AI assistant had handled 2.3 million conversations in its first month — two-thirds of the company’s customer-service chats and, by Klarna’s own estimate, the equivalent work of 700 full-time agents, with a projected $40 million profit improvement for 2024. These were company-reported figures, launched jointly with the vendor.
Fifteen months later, in May 2025, CEO Sebastian Siemiatkowski told Bloomberg the cost-cutting drive had ‘gone too far’ and that Klarna was recruiting human agents so customers would always have the option of speaking to a person. Klarna’s headcount had fallen from 5,527 (end-2022) to 3,422 (end-2024) — largely through attrition and a hiring freeze, the company said.
Automation has a quality frontier. Klarna found it by crossing it.
The 2024 numbers were the company’s own; the 2025 correction was the tell. Two board questions would have set the dial better from the start:
- Where does customer-service quality break as automation rises — and how would we know before customers tell us?
- What is the human fallback, and is it one click away everywhere?
The answer is rarely ‘maximum automation.’ It is automation scoped to where it measurably beats a human, with a person reachable everywhere else. Scoping that boundary is the work of AI automation strategy and The Proof Standard™.
Read neither number at face value.
The 2024 figures were self-reported PR, launched jointly with the vendor that supplied the model — useful as intent, not as audited fact. The 2025 ‘reversal’ was also partly narrative: Klarna’s headcount fell mostly through attrition and a hiring freeze, and AI remains central to the company. The durable lesson is about claim-verification and scope, not a verdict that AI customer service ‘failed.’
Klarna’s AI: common questions.
Did Klarna replace 700 agents with AI?
Klarna said its AI assistant did the equivalent work of 700 full-time agents in its first month — a company-reported figure announced jointly with OpenAI in February 2024, not an independently audited count. By 2025 the company was rehiring human agents, so the headline overstated how completely AI could replace the role.
Did Klarna’s AI customer service fail?
Not exactly. The AI handled high volumes, but in May 2025 the CEO said the automation push had ‘gone too far’ and that quality and the option to reach a human mattered more than the company had assumed. It was a scope correction, not a total failure — AI remains central to Klarna.
Why did Klarna start rehiring humans?
CEO Sebastian Siemiatkowski told Bloomberg in May 2025 that cost-cutting through AI had ‘gone too far’ and launched a recruitment drive so customers would always have the option of speaking to a real person. The driver was service quality and customer choice, not a failure of the technology itself.
What is the lesson for companies automating customer service?
Find where automation’s quality frontier is before you cross it, and keep a human reachable everywhere. Treat vendor and self-reported metrics as claims to verify, not facts. The right target is automation scoped to where it measurably beats a human — not the maximum the technology allows.