Bedrock and Other AWS AI Systems Power 'Help Me Decide' Shopping Tool

• By David Ramel
• 10/27/2025

Amazon has introduced a new generative AI shopping feature called "Help Me Decide," designed to streamline product selection by recommending the best option with a single tap. The new feature leverages large language models (LLMs) along with multiple Amazon Web Services (AWS) offerings, including Amazon Bedrock, Amazon OpenSearch, and Amazon SageMaker, to analyze a shopper's browsing behavior, preferences, and purchase history.

The feature expands Amazon's suite of AI-driven shopping tools, which already includes the Rufus AI assistant for answering product questions, AI-powered Shopping Guides, and the Interests feature that surfaces items aligned with user passions. Together, these systems aim to make online shopping faster and more confident for customers.

How It Works
When customers view several similar items on the Amazon Shopping app or mobile browser, a "Help Me Decide" button appears. Tapping it prompts the system to evaluate products using AI analysis of browsing and shopping data, ultimately suggesting one clear recommendation with an explanation of why it fits the shopper's needs. The system highlights key features, references customer reviews, and compares similar products. Shoppers can also view alternative options, such as a budget pick or an upgrade pick.

Amazon this week said that the feature relies on LLMs running on Bedrock as well as OpenSearch and SageMaker to correlate user activity with product data and reviews. These AWS components enable the system to "understand what you need and why," then match that understanding to the most relevant products.

AWS has been embedding its generative AI stack directly into existing services to add task-specific assistance and automation. In business intelligence, Amazon Q in QuickSight has been generally available since last year, bringing conversational analysis, narrative generation, and other "Generative BI" features to BI users inside QuickSight itself. In contact centers, Amazon Q in Connect adds an AI assistant that surfaces answers and actions in real time for agents and end customers, with recent updates letting admins choose among different LLM families in the Connect UI to optimize for speed or reasoning.

At the platform layer, AWS is expanding Bedrock to orchestrate more complex enterprise workflows. Bedrock Flows links models with guardrails, knowledge bases, agents, and other AWS services to build multi-step applications, with previews for multi-agent collaboration, multi-turn agent conversations, and long-running executions intended for larger datasets and complex jobs.