Amazon explores global product database amid AI shopping competition

Amazon has developed an internal proposal to gather information about products sold worldwide, including goods it does not stock, according to a media report citing a company document and people familiar with the matter. The initiative, called World Product Graph, aims to keep shoppers from leaving Amazon to research purchases.

Whether the project will proceed, and on what schedule, remains unclear. The document gives October 1, 2027, as an aspirational announcement date, but people familiar with the proposal said financial or other considerations could lead to changes or cancellation. Amazon said outdated brainstorming documents can present an incomplete picture and do not necessarily reflect its current plans.

The proposal envisions obtaining information through ongoing contributions from brands and retailers, alongside Amazon’s own crawling of other retail websites. The document identifies Coca-Cola and Bayer as participants and describes a trial involving Procter & Gamble. Those companies did not respond to requests for comment.

The database would also seek to improve Amazon’s shopping assistant, previously known as Rufus and renamed Alexa for Shopping in May. The document describes incorrect or incomplete answers and says richer data could enable cheaper models. Amazon disputed the document’s usage figures as significantly outdated and said its data puts the assistant’s global accuracy at roughly 97%.

People familiar with the initiative said it could also help Amazon counter competing AI shopping agents that direct customers elsewhere. CEO Andy Jassy said during the company’s third-quarter earnings call that relatively few customers were shopping through such agents and that their effect on Amazon’s advertising business had been minimal.