Amazon wants a database covering every product sold anywhere on the internet, including products it does not sell, Reuters reported.
The project is called the "World Product Graph." A product graph is a database that links items to their details, so the same product can be recognized across different sites. Amazon would make the information available through its own site. The aim, according to Reuters, is to make shoppers less likely to leave Amazon for Reddit or Google when they research a purchase, and more likely to end up on Amazon even when the product they want is sold by someone else.
AI shopping agents are part of the plan. If Amazon holds the largest stockpile of shopping data, those agents may rely on its index of products over others. That would pull people who shop through bots toward Amazon as well.
The logic is about where the shopping journey starts. Research comes before the purchase, and a lot of it happens off Amazon, in forum threads and search results. Amazon is treating that research step as territory worth owning, even when the eventual sale goes to a competitor. An agent comparing products needs a catalog to consult, and the biggest, most complete one has an edge.
The harder question is trust. A reference that exists to steer people back to the store that built it is not the same as a neutral one, and shoppers who turn to Reddit often do so because it is not selling them anything. Whether a catalog run by the world's largest online retailer can earn that kind of standing, from people or from their bots, is the part no database can settle.