Klarna says its AI assistant, powered by OpenAI, has moved from experiment to large-scale customer service tool. In one month, the Swedish payment provider says the system handled two-thirds of all customer service chats.
The figures point to a practical shift in how AI agents are being used: not only to answer simple questions, but to absorb a major share of daily support work while still leaving customers the option to speak with live agents.
A month of customer service at AI scale
According to Klarna, the AI assistant handled 2.3 million conversations in one month. The company says that volume is equivalent to the work of 700 full-time employees.
The assistant is not limited to a single country or language. Klarna says it is available in 23 markets, runs 24/7, and supports more than 35 languages. That matters because customer support often becomes harder at scale when people need help across markets, time zones, and languages.
The company says the assistant handles work ranging from multilingual customer service to managing refunds and returns. Klarna also plans to add more features in the near future.
For customers, the most direct change is speed. Klarna says people can now resolve issues in less than two minutes, compared to eleven minutes previously. That is a major operational difference if it holds across a large share of customer interactions.
Speed did not come with lower satisfaction, Klarna says
A common concern with automated customer service is that faster answers may come at the expense of quality. Klarna says that did not happen in this case. Customer satisfaction is on par with human agents, according to the company.
Klarna also says the error rate in resolving queries declined by 25 percent. In plain terms, the company is claiming that the AI assistant is not just moving conversations faster, but also reducing the share of customer issues that are resolved incorrectly.
The source figures frame the assistant as a support tool with three reported advantages:
- It handled two-thirds of all customer service chats in one month.
- It resolved customer issues in less than two minutes, compared to eleven minutes previously.
- It reduced the error rate in resolving queries by 25 percent while keeping customer satisfaction on par with human agents.
Klarna also says the assistant has improved communication with local immigrant and expat communities in all markets by supporting multiple languages. That point is important because language coverage can determine whether customers can clearly explain a problem and understand the answer.
Still, Klarna says customers can continue to interact with live agents if they prefer. That keeps a human path available for people who do not want to use the assistant or whose issue may require another route.
The job-market question is now part of the story
Klarna CEO Sebastian Siemiatkowski discussed the results on X and connected them to a broader question: what AI means for society and work. In Klarna's case, the AI agent is not directly leading to layoffs, because the company has outsourced its customer service.
Siemiatkowski wrote, "We decided to share these statistics to raise the awareness and encourage a proactive approach to the topic of AI. For decision makers worldwide to recognise this is not just "in the future", this is happening right now,"
That statement places Klarna's numbers in a wider debate. If one assistant can take on a volume of work Klarna describes as equivalent to 700 full-time employees, the implications go beyond one company's support desk.
Klarna estimates that the assistant will add $40 million to its profits by 2024. That estimate shows why companies are paying close attention to AI agents: the appeal is not only customer convenience, but also measurable business impact.
AI may change roles, not only remove them
The International Monetary Fund (IMF) estimates that AI could affect up to 40 percent of jobs worldwide, and up to 60 percent in advanced economies. The OECD points out that the impact does not necessarily have to mean job losses. It can also mean changes in job roles and the quality of work.
That distinction matters when looking at Klarna's example. The company is describing a system that can carry a large amount of customer service work, but the source also notes that customers can still choose live agents. The more realistic question may be how human support roles change when AI handles more routine, multilingual, or high-volume interactions.
Sam Altman, CEO of OpenAI, expects the impact of AI on the job market to be rapid. He is also optimistic that humanity will adapt and create "different ways to find fulfillment."
Klarna's reported results do not settle the debate over AI and employment. They do, however, provide a concrete example of the scale companies are beginning to report: millions of conversations, dozens of markets, more than 35 languages, and customer service performance that Klarna says matches human satisfaction while reducing errors.
For businesses watching AI agents, the lesson is straightforward. The relevant questions are no longer only whether an AI assistant can answer a customer. They are whether it can do so reliably, across languages, at large volume, with a clear route back to human support when customers want it.