AI Shopping Is Coming to Google

Google is looking to further integrate artificial intelligence into how people discover and buy products online, including a feature that can complete checkout on a user’s behalf.

The search company announced a set of shopping-related updates at its annual I/O developer conference Tuesday. Shopping will be a key capability of its new AI mode, which Google plans to roll out to all US users in the coming weeks after testing it in its experimental platform, Search Labs.

AI mode expands the AI-generated overviews Google provides at the top of search results and gives summarised answers to queries rather than lists of links, similar to the AI experiences offered by competitors such as ChatGPT. Shoppers will be able to ask complex questions, such as what pairs well with a particular item or what’s a good travel bag for rainy weather, and the AI will respond with answers and relevant product listings that the shopper can further refine with additional questions.

Users of Google’s payment service, Google Pay, will additionally get the ability to have AI purchase products for them. The feature, which Google will make available in the months ahead, allows users to be notified when an item they’ve selected, down to specifications such as size and colour, falls to a price they’ve set. If they click the option to buy within the notification, Google’s AI will go to the product listing, add it to cart and purchase it for the user.

Google will also begin testing a feature on Tuesday that for the first time will allow users to apply its virtual try-on technology to pictures of themselves. Where previously they could only see how products would appear on different models Google selected, now they’ll be able to upload a full-length photo of themselves to see how any of millions of apparel items in Google’s shopping listings might look on them. The option will only be available in Search Labs, but Lilian Rincon, vice president of product for Google Shopping, said the company is always evaluating whether it can roll out features it’s testing to all users.

“Over the last few years, we have really been trying to transform Google shopping with AI,” Rincon said.

Shopping has become an active new frontier in AI. In April, OpenAI announced it would launch shopping within ChatGPT, following AI company Perplexity’s release of its own shopping feature late last year.

Tech companies and fashion executives alike are hoping AI could improve online shopping by making it easier for shoppers to sift through the often-overwhelming number of options available and find the right product. Half of executives identified search and discovery as the area where generative AI held the most potential in the State of Fashion 2025 by The Business of Fashion and McKinsey.

At the same time, the volume of search traffic to retailers from AI sites is quickly growing. Adobe Analytics found that, during the 2024 holiday season, traffic from generative-AI sources grew 1,300 percent compared to the previous year, and in February, after the holiday season had ended, traffic was still up 1,200 percent compared to July 2024.

Rincon said Google sees shopping in AI mode as just another way for consumers to find and buy products, not as a replacement for Google’s traditional search and shopping options. Shoppers who are logged into Google will also see results that are more personalised based on their past activity.

“This is finally a way to be able to have a tailored conversation with Google to get at exactly what you want,” she said.

Generative AI hasn’t been without its stumbles. One of the biggest challenges for the technology, rooted in making statistically probable inferences rather than genuine understanding, is hallucinations where it provides inaccurate information. Google’s AI overviews, for instance, occasionally contain incorrect details. It raises the question of whether an AI agent could show you the wrong price for a product and then complete a purchase that you authorised at a cost you wouldn’t have accepted otherwise.

Google is being very cautious with its agentic checkout and the issue of hallucinations more broadly, according to Rincon. If the product is priced higher than a user agrees to pay, the AI wouldn’t buy it, she said. The information it uses is drawn from Google’s shopping graph, the company’s data set of more than 50 billion product listings, about 2 billion of which are refreshed every hour.

Rincon also noted that Google doesn’t share any photos users upload for virtual try-on with other services.

While some brands might welcome the growth of AI shopping, others could be worried that their products won’t be surfaced in the results users see. Rincon’s advice is the same they’ve given to brands looking to keep their items at the top of Google Shopping: Make sure all your product information is up to date on your sites and in Google’s Merchant Center.

“We’re just continuously listening to feedback from users and from merchants to bring shopping to this new era,” she said.