Klarna CEO Sebastian Siemiatkowski went all-in on AI in 2023, cutting 40% of staff and replacing them with a chatbot that handled millions of conversations. By 2025, he reversed course, bringing back human support after customer satisfaction dropped and the PR damage mounted.
Chapters
In 2023, the CEO of Klarna said AI would replace half his workforce in just a few years.
AI can already do all of the jobs that we as humans do.
Sebastian Chmiakovsky froze hiring, called AI the future, and went all in. He even bragged about how Klarna's AI chatbot was doing the work of 700 customer service reps. This AI first strategy was based on the idea that the company would get more and more efficient with fewer humans in the loop. And Klarna eventually reduced its workforce by 40%. But in 2025, Sebastian flipped that narrative on its head. He publicly stated that quality human support is the way of the future for us. So why the sudden change? First up, though, is a bit of background on Klarna. It's a Swedish fintech company that offers buy now, pay later services like Afterpay, and has over 90 million global customers. Up until 2019 or so, Klarna was profitable. But in 2020, the company started losing money, and two years later, they posted their worst ever financial year, $1 billion in losses. Sebastian was looking to cut costs wherever he could. Then, in 2023, he became obsessed with ChatGPT. He reached out to Sam Altman, asking if Klarna could test out OpenAI's tech. And there was one specific function Sebastian wanted to automate. At the time, Klarna was handling around three and a half million customer chats a month. So after integrating with some of OpenAI's products, Klarner released a chatbot. The goal was to replace as many customer service reps as possible. And Sebastian was quite bullish on this.
A lot of jobs are going to be threatened.
He'd gone as far as putting a hiring freeze in place to try and shrink the size of his workforce due.
To the implications of AI. Since September October, we have stopped recruitment.
Jump to 2024, and the results were looking promising. Within a month of the chatbot's launch, it had taken over two thirds of all customer chats. Sebastian bragged it did the work of 700 customer service reps. The resolution time of a ticket went down from 11 minutes to two, and by this point, they'd gone from 5,000 to 3,500 employees. And Sebastian planned to cut even further, to 2,000 people. He kept saying, there are so many people doing similar things. Klarna even launched an internal AI assistant called Kiki. It apparently took 2,000 inquiries a day from employees. The media ate it up. Sebastian became the poster child for AI deployment. It got to the point that by the end of 2024. At the annual financial review, Klarna's results were delivered entirely by a deep fake of the CEO.
I'm proud to announce a really strong set of results for the first nine months of the year.
23% year on year growth and a return to profitability. The chatbot alone, it saved the company $40 million in one year just by reducing support costs. But there was a problem. Things quickly started to change when 2025 came around because it turns out customers didn't love the chatbot. Reddit was buzzing with complaints. Social media lit up, calling it a filter that routes users to a human without actually solving the problem. Satisfaction scores started to drop and the number of complaints kept going up. This was probably the main reason Sebastian changed his mind on AI. In February 2025, he tweeted about getting their strategy wrong, saying, in a world of AI, nothing will be as valuable. He also changed his tune on job security. Telling reporters, my work to me is a super important part of who I am, and realizing it might become unnecessary is gloomy. Maybe that was to show some empathy and boost morale. This is because his employees had been showing up to work knowing their jobs were on the chopping block. But the other thing was timing. Klarna's rumored to go public very soon, and so before the ipo, they probably wanted to avoid dealing with a PR campaign over AI taking jobs. But overall, Sebastian admitted they had overindex on cutting costs. And in May 2025, he openly acknowledged that this resulted in a lower quality service. This is where the company shifted its focus away from short term productivity gains and cost savings. Klarna is now chasing three things. One, fewer customer complaints, Two, more issues resolved, and three, better customer experience, even if it means more humans on payroll. That's a pretty big shift from a CEO who once claimed AI would replace us all. The new narrative is all about balance. Sebastian calls it a hybrid approach. It's where machines do what they do best, which is things like speed and scale and humans do what we do best, which is empathy, taste, judgment and complex problem solving. AI working with human oversight to deliver the best customer experience possible. And so, under this new strategy, Klarna is hiring again. Sebastian has abandoned those plans to half his workforce using AI. Instead of the company is looking for more contractors, graduates and even loyal users to work for them. It's also piloting an Uber type of setup in Sweden. This is where workers choose their own schedules and can work remotely. It's a clear signal Klarna is committed to keeping humans in the loop. But it's not just Klarna facing this kind of backlash. Duolingo just got roasted on TikTok about its AI first approach. This is after laying off 10% of its translation contractors. Then Shopify CEO is also under fire. He said no new hires unless you prove AI can't do the job. Company CEO is encouraging its employees to weigh their need for workers against the talents of AI. And that's what blows my mind, because Klarna did the exact same thing two years ago and it hasn't worked out. The financial services company net loss more than doubled in the first quarter, totaling $99 million. It's a public case study on what happens when the hype wears off in real time. And it might be a glimpse into what's next for almost AI First Company. Because when it comes to any new technology, there's one important question to ask. Are we focusing on what actually matters to customers.
