What Amazon’s Agentic Shopping Assistant Is
Amazon’s Agentic Shopping Assistant is an AI shopping assistant framework that lets retailers build branded, conversational helpers that guide customers from product discovery to purchase across web or app storefronts. It combines generative AI, search, and retail data so shoppers can describe what they want in natural language instead of typing keywords into a search bar. The system runs on Amazon AWS retail infrastructure, powered by services such as Amazon Bedrock for generative AI, AgentCore for managing AI agents, and OpenSearch for search and retrieval. Amazon says this packaged foundation lets retailers deploy AI customer service experiences “in weeks” instead of spending years on custom development. The assistant originates from the Alexa for Shopping work on Amazon.com, validated on billions of shopping interactions and more than 300 million users, and is now repurposed as a retail technology solution that can be tailored to any brand’s catalog and voice.
From Alexa for Shopping to a New AWS Revenue Stream
Amazon is turning its own AI shopping experiments into a business line for Amazon AWS retail customers. The Agentic Shopping Assistant distills lessons from Alexa for Shopping, the company’s in-house agent that merges Rufus and Alexa+ into a conversational shopping experience. Amazon reports that its AI assistant was used by more than 300 million customers last year and contributed to nearly US$12 billion (approx. RM55.68 billion) in incremental sales, a clear sign that AI customer service can move the needle on revenue. Now, Amazon is packaging that know-how as starter code, reference architecture, and expert guidance via the AWS Generative AI Innovation Center and partner system integrators. The strategy is straightforward: make AWS the default infrastructure backbone for retail technology solutions, so that when retailers adopt AI shopping assistants, they also deepen their commitment to Amazon’s cloud tools, models, and agent orchestration layer.
Kate Spade’s AI Gift Concierge: A Case Study
Kate Spade New York is one of the first brands to implement Amazon’s Agentic Shopping Assistant, launching an AI Gift Concierge after about 2.5 months of testing. Built on top of the AWS stack and powered by Anthropic’s Haiku 4.5 model, the assistant focuses on gift buying, a process Amazon says stresses 53% of shoppers. Customers describe the occasion, recipient, and style preferences, and the AI shopping assistant recommends gifts that match Kate Spade’s catalog and brand identity. Tapestry executives describe it as a conversational experience that feels less like search and more like “talking to someone who knows the brand and knows how to give a great gift.” Amazon calls it the first production-ready retail AI assistant built with AgentCore, using Amazon Bedrock for observability, authentication, and evaluation, and positioning it as proof that a polished AI customer service agent can be launched in a matter of weeks.
Why Retailers Care: Experience, Conversion, and Control
For retailers, the appeal of Amazon’s Agentic Shopping Assistant lies in faster time-to-market and immediate access to patterns learned from billions of Amazon.com interactions. Amazon says conversational shopping sessions can generate conversion rates 3.5 times higher than traditional keyword-based search, which makes AI shopping assistants attractive as both a customer experience and revenue play. Each deployment can be customised to a retailer’s catalogue, customer base, shopping environment, and brand voice, letting merchants keep control over their data, merchandising logic, and customer relationships. At the same time, they tap into Amazon’s mature tooling for search, retrieval, and AI agent operations. This balance—shared infrastructure plus proprietary insight—helps retailers modernise AI customer service without handing over their competitive advantages, and positions AWS as the backbone for future retail technology solutions that blend human-like guidance with data-driven product discovery.
Signals of a Broader Shift in Retail AI
Amazon’s move to commercialise its AI shopping stack through AWS marks a shift from AI as an internal tool to AI as shared infrastructure for the wider retail industry. Early adopters such as Kate Spade show how brands can narrow the scope—gift shopping, in this case—while still benefiting from Amazon’s agentic commerce expertise. Inside Tapestry, AI is also used through Mira, an internal platform built with Amazon Bedrock that helps employees manage assortments and inventory by surfacing insights in seconds to minutes. Together, these efforts hint at an industry where AI customer service agents face shoppers, while separate AI tools support employees behind the scenes. As more retailers test the Agentic Shopping Assistant, Amazon’s bet is that AWS becomes the default platform for both sides of that equation, anchoring the next wave of AI-powered e-commerce innovation.
