Amazon’s AI shopping tool learns your buying habits to recommend products

Amazon has launched “Help Me Decide,” an AI-powered shopping tool that analyzes your purchase history and preferences to recommend products tailored to your specific needs. The feature uses large language models and AWS cloud computing services to cut through decision paralysis by providing personalized product recommendations with explanations for why each item was chosen.

How it works: Help Me Decide leverages machine learning to analyze multiple data points and create targeted recommendations.

• The tool examines products you’ve viewed, your order history, and personal shopping patterns to understand your specific needs.
• For example, if you’ve browsed adult and kids’ sleeping bags for cold weather plus car camping stoves, it recognizes you’re planning a family camping trip and might recommend a four-person tent.
• Amazon uses large language models combined with AWS services like Amazon Bedrock, Amazon OpenSearch, and Amazon SageMaker to power the analysis.
• Each recommendation includes an explanation of why that particular product was selected for you.

What you should know: The feature appears automatically in Amazon’s mobile app and browser after you’ve viewed several similar items.

• You can access it directly by tapping “Keep shopping for” at the top of the homepage.
• Beyond the main recommendation, the tool also provides “upgrade” and “budget” alternatives.
• Amazon hasn’t disclosed how the AI selects which products to recommend or whether paid placements will influence suggestions.

The bigger picture: Help Me Decide represents Amazon’s latest AI shopping innovation, joining a growing suite of personalized tools.

• The company’s AI-powered “Interests” feature scans for new products matching your preferences and sends notifications.
• “Shopping Guides” compiles expert guidance and product recommendations to simplify research.
• Rufus, Amazon’s AI shopping assistant, answers product questions in real time and can access your previous orders.