Rufus worked. That is what makes this move interesting.
Amazon's Rufus AI shopping assistant, launched in early 2024, generated nearly $12 billion in additional sales across 2025 and was used by over 300 million customers that year. People who talked to Rufus during a shopping session were 60% more likely to complete a purchase compared to those who did not. Monthly users were growing at over 140% year-on-year. By almost every measure, it was one of the most commercially effective AI deployments in retail history.
And Amazon retired it anyway. The reason is scale.
Alexa for Shopping replaces Rufus inside Amazon's main search bar, on the Amazon app, and on Echo Show devices. It is not a side tool anymore. It is the front door. When you type a question into Amazon, you are now talking to an AI assistant that combines everything Rufus knew about products with everything Alexa knows about you personally: your conversations at home, your purchase history, your preferences shared across devices, and your activity across Amazon's broader services including Kindle and Prime Video.
The data connection is the entire point. Amazon has explicitly said that what you say to Alexa on your Echo in the kitchen informs what you see when you open Amazon on your phone the next day — and vice versa. If you asked Alexa to help plan a child's science fair project, Alexa for Shopping will remember that when you later open Amazon and ask for supply suggestions. This cross-device memory is something no outside AI tool can replicate, and it is Amazon's clearest advantage over ChatGPT, Google, or any other company trying to build AI shopping tools.
For the hundreds of millions of people who shop on Amazon, the practical experience will feel like a smarter search bar. Instead of typing "skincare for men" and getting a wall of results, you can describe your situation and get a guided answer. The assistant can also track prices, alert you to deals, reorder household essentials, and in some cases make purchases for you automatically at a price you set in advance.
But for brands and sellers on Amazon, the shift carries a real risk. Amazon's advertising business generates over $60 billion a year, most of it from sellers paying to appear at the top of search results. When an AI assistant answers a shopper's question directly, it decides what gets recommended — and paid placement in a traditional list does not automatically transfer to an AI-guided recommendation. Amazon says ads will still appear where they are relevant, but the definition of relevant is now controlled by Amazon's own AI, not by an open auction. The platform is essentially building a new layer between sellers and customers, and the rules of that layer are still being written.
There is also a data question that most consumers will not think about but probably should. As of March 2025, Amazon removed the option for users to prevent their Alexa voice recordings from being stored. All interactions are now processed and retained in Amazon's systems and used to train its AI. Alexa for Shopping deepens this further by explicitly linking what you say out loud at home to what you buy online. The Alexa Privacy Dashboard lets users review and manage some of this, but the default position is full data collection — and opting out in any meaningful way means opting out of the product itself.
Amazon is making a large bet: that consumers will prefer a deeply personalized shopping assistant over an open search box, and that the convenience will outweigh any discomfort with how much the system knows about them. Given what Rufus achieved with far less personal data, the bet is not unreasonable. But it also means Amazon is becoming something qualitatively different from a store. It is becoming an AI that happens to sell things — and that has been listening to your household conversations for years.