Every week, it seems, another headline announces that AI can now do the work of dozens, hundreds or thousands of people. For a small business owner juggling wages, rising costs and a stretched team, the temptation is obvious: if a chatbot can handle customer enquiries, why pay people to do it?
One of the most widely discussed examples of a company acting on that temptation, and then partly reversing course, is the Swedish payments company Klarna. Its story is useful precisely because it shows what happens when cost becomes the main measure of success, rather than offering a simple tale of AI succeeding or failing.
For small businesses, where every customer relationship counts, the lessons are especially relevant.
One Like insight: AI can cut the cost of service. Only people can protect the quality of the relationship.
The Klarna story
In February 2024, Klarna announced that its new AI assistant, built with OpenAI, had handled 2.3 million customer conversations in its first month, around two-thirds of its customer-service chats. The company said the assistant was doing the equivalent work of 700 full-time agents, with customer satisfaction on par with human agents, faster resolution times and fewer repeat enquiries. It projected a significant profit improvement as a result. Klarna had already paused most hiring and was reducing its headcount.
By May 2025, the tone had changed. Klarna’s chief executive, Sebastian Siemiatkowski, told Bloomberg that the company would hire human customer-service staff again. He explained that cost had become too predominant an evaluation factor and that the result was lower quality. The company said it would invest in human support and make sure customers could always reach a person, while continuing to use AI.
Klarna has described its approach as combining AI and people rather than abandoning automation, and some of its original efficiency gains were real. Commentators have debated how much of the reversal was about service quality and how much about other factors. But the story has become a widely cited reminder that headline claims about AI replacing people deserve scrutiny, and that customers notice when service quality slips.
Balancing AI and human service
For small businesses, Klarna’s experience points to a balanced approach rather than an all-or-nothing choice.
Let AI handle the simple, repetitive questions. Opening hours, order tracking, booking changes and common “how do I” questions are often well suited to AI, freeing people for work that needs judgement and empathy.
Always offer a clear route to a person. Customers who are upset, confused or dealing with something important should be able to reach a human quickly and easily. As we explored in Make It Easy, making customers work hard is what damages loyalty most.
Measure quality, not just cost. Track customer satisfaction, complaints, repeat contacts and whether problems actually get solved, not only how much money the AI saves.
Use AI to support your team. As AI Can Help Your Newest Staff Work Like Your Best showed, some of the clearest gains come from AI helping people, rather than replacing them.
Remember your advantage. For small businesses, personal service is often a key reason customers choose them over larger competitors. Automating it away entirely can remove the very thing that makes you different.
[Infographic placement] Where AI fits in customer service. Two columns. Good for AI: opening hours, order status, simple FAQs, booking changes, first-line triage. Keep human: complaints, complex problems, upset customers, important or emotional decisions, relationship-building. Beneath, a timeline: Feb 2024 — AI assistant handles 2.3m chats, “work of 700 agents”; May 2025 — Klarna resumes hiring human agents to protect quality. Caption: Klarna press release (2024); Bloomberg and CX Dive (2025).
Getting the balance right
If you’re considering AI for customer service, start small. Introduce it for one narrow job, such as answering frequently asked questions out of hours, with a clear option to leave a message or speak to a person. Measure both the savings and the quality. Read the conversations regularly. Ask customers how they found it. Expand only where the evidence shows customers are well served.
A few questions can help:
- Which customer questions are simple and repetitive enough for AI?
- How easily can a customer reach a person when they need one?
- Are we measuring service quality, or only cost?
- Is personal service part of why customers choose us, and how will we protect it?
Which brings us back to the small business owner reading another headline about AI replacing hundreds of jobs. The lesson from Klarna isn’t to avoid AI. It’s to use it where it genuinely helps customers and staff, keep people where judgement and empathy matter, and judge success by the quality of the relationship, not just the size of the saving.
Sources
- Klarna, “Klarna AI assistant handles two-thirds of customer service chats in its first month”, press release, February 2024.
- Bloomberg, “Klarna Turns From AI to Real Person Customer Service”, 8 May 2025, as reported by CX Dive.
