Operational Excellence (OPEX) Insight – Tuesday - November 17, 2026: Klarna Accelerates on Its AI Assistant
Klarna Accelerates on Its AI Assistant
WELCOME TO THE NEW BIZINSIDER - YOUR NEW HOME FOR KNOWLEDGE, AI & PROFESSIONAL GROWTH!
BizInsider has officially moved to its upgraded home at https://bizinsider.co/, bringing you more than just articles! Enjoy bilingual insights, Ask AI, podcasts, KnowLay courses, Career Insights, exclusive webinars, and expanded membership benefits. Already a member? Simply use your existing registered email to access the new platform, maintain your membership, or upgrade to unlock even greater value. Don’t miss our exclusive 20% welcome offer for existing members!
Make the move today: https://bizinsider.co/
On November 17, 2026, Klarna, the Swedish fintech company known for buy now pay later, reported third quarter results with revenue above forecast and a new step in its artificial intelligence shopping assistant. Quarterly revenue reached about 1.01 billion dollars, up 12 percent from the same period a year earlier. Total transaction value reached 36.1 billion dollars, up 10 percent, of which the United States market grew 24 percent and the rest outside the United States grew 7 percent. Active customers reached 122 million worldwide. Merchant partners reached 1.35 million, up 59 percent from a year earlier. Subscription revenue reached 43 million dollars, up 530 percent, with the number of paying subscribers tripling. Card payment volume reached 4.5 billion dollars, more than double the prior year. According to news reports, the stock at one point rose more than 5 percent after the release was published, before pulling back during the session.
What makes this report worth reading closely does not lie in the revenue figure, but in what came with it. On the same day, Klarna introduced an upgraded version of its artificial intelligence shopping assistant. This assistant no longer only answers questions, but can place orders, track deliveries, and handle returns. The tool is running in a trial with a small group of users through WhatsApp, iMessage, and the Klarna app, while preregistration is open. The company founder and chief executive called this an opportunity in agentic commerce, where a new feature and a new experience are added to the shopping journey. Klarna has not disclosed the number of orders the assistant handled during the trial.
A good quarterly report alongside a new feature sounds like simple good news. But for someone working in operations, the value of this news lies elsewhere. Klarna is one of the few companies that has publicly shown the whole path of using artificial intelligence to change how it serves customers, including both the successful part and the part that had to be adjusted. When a company both shows off a new feature and has a full implementation history behind it, the reader gains a valuable chance to learn, not from advertising, but from how an organization actually runs an idea from the moment it begins until the moment it must be corrected.
For that reason this article will not stop at retelling the numbers. The article will use one operational tool to read the entire Klarna story, from the new feature to the lessons already available. That tool is A3, a way of solving a problem presented compactly on a single page. The name sounds unfamiliar, but the method is close to anyone who has ever had to handle a hard problem without wanting to lose direction. A3 is one of the least expensive and most common tools in the operational toolkit, and it fits especially well with the situation of a business weighing whether to bring a new technology into its machinery.
Before entering the tool, it helps to look back at Klarna‘s artificial intelligence path, because that is the part that offers the most lessons. In early 2024, Klarna announced a result that drew industry wide attention. The company’s artificial intelligence assistant, built together with a large technology partner, in its first month handled 2.3 million conversations with customers, equivalent to about two thirds of all chats. This workload was described by the company as equivalent to 700 full time agents. The time to resolve a request fell from 11 minutes to under 2 minutes. The number of times a customer had to ask again about the same matter fell 25 percent. The company projected a profit improvement of about 40 million dollars for 2024. It should be stated clearly that the 40 million dollar figure was a forecast at that time, not a result that had been fully measured.
The story did not stop there, and this is the important part. According to information the company itself disclosed, by the first quarter report of 2025, the cost of serving customers per transaction had fallen 40 percent over two years, while customer satisfaction was held level. By the third quarter report of 2025, the assistant was said to be doing a workload equivalent to 853 full time agents and had saved about 60 million dollars. Looking at this sequence of numbers, it seems everything only moved in one direction, namely more and more work automated and more and more saved.
But in mid 2025, the head of Klarna himself publicly adjusted the view. He told a news agency that cost had become an evaluation criterion that overwhelmed too much in how the customer service department was organized. After that statement, the company began a trial program to rehire remote staff with flexible hours. The artificial intelligence assistant remained and still handled about two thirds of requests, but people returned to take on the more important situations. In other words, a company once seen as a model of replacing people with machines had itself shifted toward a model of cooperation between people and machines.
This is why the Klarna story is worth more than a financial news item. It shows a complete life cycle of an operational decision, including an initial phase of excitement with beautiful numbers, a phase of expansion, then a phase of realizing that a single criterion had been pushed too far, and finally a phase of adjustment. Most businesses do not get to witness such a full life cycle at another company, because people usually show off only the first part and hide the correction part. Having both parts in hand is an ideal condition to practice a problem solving tool.
Plus
This post is for members
Subscribe to read the full post and more in-depth analysis from BizInsider.








