Amazon brings its ‘Rufus’ AI shopping assistant to India

Amazon has been enhancing its shopping platform with AI technologies for some time, offering features such as summarized reviews and personalized product recommendations. Now, the company has rolled out a new AI chatbot called Rufus, which is currently in beta testing and accessible to select Android and iOS users in India.

Rufus allows users to ask questions about specific products, such as “What should I keep in mind when buying a washing machine?” or “Should I opt for a fitness band or a smartwatch?” The chatbot is designed to provide helpful responses and suggest products based on individual interests.

Trained on Amazon’s vast dataset, which includes customer reviews and product catalogues as well as relevant public data, Rufus can handle a range of natural language inquiries like “Can you recommend gifts for children under 5?” or “How do various coffee makers compare?” This makes Rufus a valuable shopping companion, as it integrates Amazon’s extensive inventory with broader online information to assist customers in making informed purchase decisions.

To start using Rufus, shoppers in India and other newly added markets need to update their Amazon Shopping app to the latest version. They can then access the chatbot by tapping an icon in the bottom right corner of the app, which opens a familiar chat interface. Users can expand the chat dialogue box to see responses to their questions, choose from suggested queries, and ask follow-up questions as needed. If they prefer to return to the standard search results, they can easily do so by swiping down to minimize the chat.

Rufus was initially trialed in the U.S. in February before its official launch five months later. Amazon is also working on additional generative AI tools aimed at helping sellers improve their listings by generating product descriptions and titles.

Furthermore, the company has invested $230 million to support generative AI startups. However, Amazon cautions that Rufus is still in its early stages and may not always deliver accurate responses, a common limitation among generative AI applications.