Klarna: an AI assistant handling two-thirds of customer service chats
Klarna says its OpenAI-powered assistant handled 2.3 million chats in its first month, two-thirds of service chats. In 2025 it began hiring human agents again.
Klarna handed the first line of customer service to AI. According to the company, its assistant handled 2.3 million conversations in its first month, about two-thirds of all customer service chats and the equivalent of 700 full-time agents.
At a glance
| Item | Detail |
|---|---|
| Company | Klarna |
| Industry | Fintech |
| Use cases | Customer service automation |
| Reported results | 2.3 million chats in month one; resolution time down from 11 to under 2 minutes; repeat inquiries down 25%; projected $40 million profit improvement in 2024. Klarna later rebuilt human support after quality concerns |
Case details
Background
In its press release Klarna says it serves 150 million consumers worldwide. Service requests cover refunds, returns, payments, cancellations and disputes, and are high in volume and highly repetitive.
What they did
Klarna launched an OpenAI-powered customer service assistant inside the Klarna app. It is available around the clock in 23 markets and in more than 35 languages. It handles refunds, returns, payment issues, cancellations, disputes and invoice errors, and can update balances and repayment plans and explain spending limits. Customers can still choose to talk to a human agent.
Results and data
| Metric | As stated by Klarna (February 2024) |
|---|---|
| Conversations in the first month | 2.3 million |
| Share of service chats | About two-thirds |
| Equivalent workload | About 700 full-time agents |
| Customer satisfaction | On par with human agents |
| Time to resolve | From 11 minutes to under 2 |
| Repeat inquiries | Down 25% |
| 2024 profit improvement | Projected at about $40 million |
In May 2025, CEO Sebastian Siemiatkowski told Bloomberg that cost had been weighted too heavily in past evaluations, and that AI support was cheaper but lower in quality. Klarna began recruiting remote human agents, aiming to make sure customers can always reach a person.
What to take from it
- Start with frequent, standardized questions. Refunds, returns and payments are high-volume and rule-based, which makes them the natural first job for AI.
- Keep a human channel. Klarna allowed handoff to people from launch and later strengthened human support again; a human fallback is not optional.
Limits
- The first-month figures are self-reported by Klarna and not audited. “700 agents” is an equivalence, not actual layoffs, and the $40 million is a projection.
- The CEO publicly acknowledged quality problems in May 2025, so first-month data alone overstates the long-term effect.
- Businesses with many long-tail or complex inquiries should not expect the same automation rate.
Sources
Original source: Klarna press release · Klarna AI assistant handles two-thirds of customer service chats in its first month Additional source: Entrepreneur · Klarna Is Hiring Customer Service Agents After AI Couldn’t Cut It on Calls
This is an industry research entry: a publicly documented third-party AI case, not a Moments client case. For projects Moments delivered itself, see client cases.
Related
- IKEA Billie: an AI service bot and 8,500 agents retrained as designers — Ingka Group's AI bot Billie handled about 47% of IKEA call center queries over two years, while 8,500 agents were retrained as remote interior design advisers.
- JPMorgan COIN: machine learning for commercial loan agreements — Bloomberg reported in 2017 that JPMorgan's COIN reviews commercial loan agreements in seconds, work that took lawyers and loan officers 360,000 hours a year.