Meta is testing a shopping research feature inside of its AI chatbot, enabling users to search for product recommendations and receive a carousel of product images that include captions with information about the brand, website, and price. Meta also offers a brief explanation of its recommendations as a bullet-list below the carousel.
The recommendations are personalized to each shopper, based on the obscene amount of information that Meta has on its users, including their gender and location. For example, when asked to find puffer jackets, Meta AI's response referenced the author's location in New York and offered options for women's puffers.
Bloomberg notes that there is no checkout or payment option within the chatbot, but users can click on the product links to view the item on the merchant's website.
As we learned from OpenAI's experience with agentic commerce, Meta isn't missing out on anything by not offering it!
Checkout is a low margin game. Advertising is where the profits are at, and that's what Meta excels at. It appears that they are developing a shopping discovery tool that can cater to the trove of data they have on their users to personalize results, as opposed to attempting to control the whole purchase flow. Seems a lot more manageable, while fitting in perfectly with Meta's existing business model.
It's actually surprising that Meta hasn't offered a more robust product search for the past decade, long before AI and agentic commerce were household names. When Meta launched Facebook Marketplace in 2016, it was the perfect opportunity to build a product search and recommendation engine that intermixed Marketplace results with products from the web, while cementing Facebook as destination for product discovery. No time like the present though.
