A mobile phone in hand, a few minutes left until the bus and a quick question: which running shoes can withstand rain, what do you give a 13-year-old who already seems to have everything and which air fryer is actually worth the money? More and more often, the purchase starts there. Not in the store and not in a long Google search but in an AI chat that has already sorted, compared and suggested.

Image: AI

In the past, the path to a purchase was often quite chaotic: you saw something, clicked on, compared too much and ended up with more options than you had originally. Now it often starts with a straightforward question. ”I need a winter jacket that can handle slush. I’m looking for a gift for someone who likes golf but already has all the equipment. I want headphones under 2,000 kronor that stay on when I run.”
AI has crept in right there, in those little moments that previously consisted mostly of care. On the couch late at night when the choice of coffee machine suddenly feels bigger than it should be. In the fitting room when two pairs of jeans have already been tested and patience is starting to run out. In the parking lot outside the grocery store when someone in the car asks what is really needed for the tacos and no one really knows if it is cilantro, lime or just some kind of order in life.

In Adyen's Swedish retail barometer, 31 percent of Swedes said they use AI when shopping. Among Generation Z, the youngest adult consumers, the proportion was even higher: 56 percent said they use AI in connection with purchases. Internationally, development has been even faster in the path to purchase itself.

Adobe reported in August that traffic from generative AI to retail sites in the US had increased sharply compared to the previous year. In January, the company also showed that such traffic during the holiday shopping season had not only grown, but more often than before led to purchases. Walmart launched the Sparky AI tool in its app in the summer of 2025, where customers can get
help find items, get reviews summarized, and put together purchases based on everyday situations like a barbecue, a kids' party, or the weekly grocery shopping. When Walmart reported its quarterly results in February, the company said that about half of its app users had used Sparky, and that those customers placed larger orders on average.

Amazon has done something similar with Rufus, which answers questions about products, compares models and helps users narrow down their choices. Amazon's own materials show that Rufus was used by over 300 million customers in 2025. Search is also changing. When OpenAI rolled out new shopping features in ChatGPT, it became possible to get product suggestions collected in a single answer, with images,
reviews and links. Now there are also functions where the user can in some cases proceed to the actual purchase within the chat. This shortens the path between question and product. Fewer tabs, fewer detours, less time
trying to understand what really distinguishes model A from model B. For those who trade, the change is noticeable
quite concrete. A family with children can ask for five weekday dinners within a certain budget and get a shopping list in a few seconds. Someone standing in a fitting room can ask for similar garments in other price ranges.

Someone going to a housewarming party can ask for gift suggestions that don't feel generic and avoid giving away yet another vase, yet another serving platter, yet another thing that's already in a cupboard somewhere. The AI then functions less as a search engine and more as a first selection. Klarna is a Swedish example of how the infrastructure behind AI shopping is now being expanded. Last fall, the company announced support for Google's solution for agent payments. This is technology that makes it possible for AI services to not only provide advice but also help the user further towards the actual purchase. In December, Klarna also launched an open product system that, according to the company, collects over 100 million products and 400 million prices. This shows where retail is heading. AI should not only suggest what you can buy, but also find the product and shorten the path to payment. At the same time, another question follows: why are these products highlighted first? When AI summarizes reviews and sorts for the user, it also affects the selection. It then becomes important to understand what is advertising, what is a recommendation and what data is used.

The Swedish Consumer Agency is clear that hidden marketing is prohibited, and the Swedish Data Protection Authority has pointed out that generative AI almost always involves the processing of personal data. But one thing is certain; AI shopping has become part of everyday life. It is found in completely ordinary environments, between the escalator and the checkout line, between the menu and dinner, between the impulse and the receipt. It has moved into the purchase before the store staff has even had time to say hello.