Amazon weighs building a global index of online products
Amazon is considering a database of products sold across the internet, including items it does not carry, in a bid to keep shoppers’ research—and potentially AI shopping agents—within its ecosystem, Reuters reported, citing an internal project document.

The proposed “World Product Graph” would make product information available through Amazon’s site. The goal, according to Reuters, is to reduce the need for shoppers to turn to Google or Reddit for research and to draw customers to Amazon even when they are looking for items sold elsewhere. A broad product index could also appeal to AI agents seeking shopping information.
Amazon has already begun showing links to other retailers’ products through its “shop direct” feature, launched last year. The company also offers an AI shopping assistant, Alexa for Shopping, previously called Rufus. Reuters reported that an internal document described problems with the assistant, saying, “Many Rufus answers are factually wrong or incomplete.” It also said running Amazon searches through the AI is expensive. The proposed data collection is intended to improve accuracy and allow the system to use cheaper models.
The project’s status remains uncertain. The document reportedly says it depends on discovering how brands organize products on their own websites, potentially requiring Amazon to crawl other retailers’ sites. Reuters said it was unclear “if or when this plan may be implemented,” though the document included an October 1st, 2027 release date.
Amazon told Reuters: “Draft, outdated documents often paint an incomplete picture and the existence of a brainstorming document such as this one doesn’t necessarily reflect our current plans.”
Comments (0)
No comments yet. Be the first!