Klarna is moving away from full automation: AI is no longer replacing humans in customer support
Swedish fintech company Klarna is bringing back real support operators — the CEO admitted that excessive reliance on artificial intelligence has reduced the quality of service.
Swedish fintech company Klarna is bringing back real support operators — the CEO admitted that excessive reliance on artificial intelligence has reduced the quality of service.
According to Bloomberg, Klarna, one of the most high-profile examples of AI-automated customer support, is changing course. The company’s CEO, Sebastian Siemiatkowski, said that due to excessive automation, the service lost quality. Now Klarna is testing a hybrid model: live operators will work in a freelance format — «like in Uber.»
The pilot already has two agents, but the company plans to expand their number. The main goal is to bring back the live chat option while maintaining the benefits of AI.
What happened: Klarna reconsiders its bet on AI in support
In 2023, Klarna actively reported on the implementation of AI in customer service, which aimed to replace thousands of agents. The company became one of the first fintechs to publicly announce its close cooperation with OpenAI and the use of their models in its daily work.
At the same time, Siemiatkowski said that chatbots handle up to 75% of customer calls in 35 languages, and that AI is already «doing the work of 700 support agents.» This statement immediately caused a wave of reactions in the market: shares of call center provider Teleperformance collapsed at the time, and investors began to prepare for the mass automation of the industry.
However, already in May 2025, the CEO admitted that focusing exclusively on saving money through automation was a mistake.
Why automation didn’t work: a quality problem
Klarna is not abandoning AI completely — on the contrary, the company continues to develop AI assistants, but not as a complete replacement for staff, but as a supplement.
«From a brand and customer trust perspective, it’s critical to leave the opportunity to communicate with a person,» Siemiatkowski said.
The reason for the strategy review was the decline in the quality of support. The rapid implementation of AI allowed to reduce costs, but did not ensure the expected customer satisfaction. The company recognized that users react negatively to overly automated interaction.
Expert Comment: Corporate Automation is Not a Sprint, but a Marathon
AI enthusiast, CEO at DevRain, and author of the book «ChatGPT, DALL·E, Midjourney: AI and Creativity» Oleksandr Krakovetsky commented on the situation on his Facebook page as follows:
«I myself often used the Klarna case in my speeches as an example of fast and effective automation. I always added that their numbers looked too good for the models of that period. And that this was most likely partly marketing,» the expert notes.
In his opinion, corporate efficiency from AI does not come instantly:
The business processes of a large organization are not always clearly described.
The data does not cover all edge cases and scenarios.
Often the processes themselves are artificially complicated.
Therefore, before implementing AI, companies must go through the stage of unifying processes, documentation, and preparing high-quality data sets. And this is not a matter of a few months.
The expert also emphasizes: «AI is often used as marketing: look how innovative we are. This is not bad, but it distorts the idea of the real effectiveness of the technology.»
What’s next: a hybrid model instead of full automation
After a valuation plunged from $45 billion to $6.7 billion in 2022, Klarna was forced to cut costs. That’s when the company completely froze hiring and focused on AI. But after restoring some of its capitalization (expected pre-IPO valuation of $15+ billion), the company is once again balancing innovation and service quality.
Klarna plans to continue using AI as a support tool, but not as a stand-alone replacement for humans. The new hybrid approach is an attempt to find a balance between scalability and customer trust.
AI-first as a concept is just beginning to take hold in business. Companies will have to rethink the architecture of their processes to get the most out of AI—otherwise, they will only multiply bugs instead of efficiency.
We will also remind you that the Kyiv International Institute of Sociology (KIIS) recently conducted a study on how Ukrainians use AI. According to its results, 64% of the population does not use artificial intelligence at all.
Як нейромережі бачать вільну та незалежну Україну? Тест dev.ua
Нейронні мережі для генерації зображень бачать світ по-своєму, їхню логіку зрозуміти часом зовсім неможливо. Але таки хочеться. На честь Дня Незалежності України редакція dev.ua вирішила провести невеликий експеримент.
Ми задали чотирьом різним нейронним мережам п’ять однакових запитів: «прапор України», «День Незалежності України», «український Крим», «перемога України» та «українці». Отриманими результатами ми ділимося з вами нижче.
У TikTok тепер можна генерувати фон за допомогою нейромережі. Ми протестували її та ділимося результатами
У TikTok з’явилася нова функція «Розумний фон». З її допомогою як фон для тіктоків можна підставляти згенеровані нейромережею зображення. Редакція dev.ua протестувала цю технологію і ділиться своїми враженнями.