Amazon announced a series of upgrades to Rufus, its AI-powered shopping assistant.
The assistant now carries stronger predictive capabilities and search features.
Hyper-personalised Product Recommendations
Rufus now remembers users’ previous shopping activity, which it utilises to generate more personalised recommendations. For example, if a user asks about gifts to buy their children, the bot will remember products they had purchased in the past and apply that to its product recommendations.
Users can also ask the assistant to re-purchase products they had before, such as replicating last week’s grocery shop. The AI assistant uses reasoning capabilities to automatically replace out of stock items with suitable replacements.
Users can make modifications to the bot’s understanding of their preferences through natural language prompts.
Advanced Visual and Text Search
Amazon also released upgrades to Rufus’ search capabilities. This includes specialised searches based on activity or purchase, such as holiday party decorations or back-to-school necessities.
The tool can also pull from influencer storefronts, where users can buy the recommended products directly by selecting “Buy for Me” or “Shop Direct,” which opens a new tab onto the product distributor’s website.
New visual search features allow for more seamless purchasing. iOS users can upload a photo of a grocery list, and ask Rufus to purchase the items for them. Amazon stated these features will advance to the point where users can upload a photo and ask a specific query such as “how do I remove a coffee stain from this rug?” – to which Rufus will generate relevant cleaning products.
The bot can also generate product recommendations based on a product photo.
Continuous Price Tracking
Price tracking is another major new feature included in Rufus. Users can view product price history from the past 30 and 90 days to evaluate the best time to buy. Rufus can send notifications once products reach a low price point, and can then automatically purchase the item.
The assistant can filter products by price, such as “Mens jeans under $40.”
Rufus’ customer service abilities have also been improved, helping users access tracking updates, returns and refund information, and billing information at any time.
AI on Amazon’s Online Storefront
The tool answers questions using natural language understanding.This includes responding to trip planning questions like “What’s the weather in Rome in April” and product questions like “What’s the battery life on this product?”
Rufus is built on Amazon Bedrock technology, Amazon’s agentic AI service. The tool leverages large language models (LLMs) like Anthropic’s Claude Sonnet and Amazon Nova to gain knowledge on customer preferences.
Rufus also gathers product insights from sources like The New York Times, USA Today, and Vogue.
The shopping tool is becoming increasingly popular. Rufus’ average users and interactions grew 149 percent and 210 percent this year, respectively.
“We’ve built and keep improving our next-gen in-store AI assistant to be a knowledgeable shopping expert right at your fingertips – one that knows you and your preferences, can help you discover products, compare options and prices, make informed buying decisions, and even take action to get things done faster for you.” said Rajiv Mehta, vice president of Search and Conversational Shopping, Amazon.
