From Alexa for Shopping to Agentic Shopping Assistant on AWS
Amazon’s Agentic Shopping Assistant is an ecommerce AI technology delivered through AWS that lets retailers create branded AI shopping agents, using Amazon’s own conversational and search stack without rebuilding it from scratch. The service packages architecture, starter code and expert support so retailers can deploy customised retail AI assistants on their websites and apps in weeks instead of years. It is built on Amazon Bedrock for generative AI, AgentCore to operate AI agents, and OpenSearch for search and retrieval, all validated on billions of real shopping interactions from Amazon.com. Amazon says more than 300 million customers used its internal AI shopping assistant last year, generating nearly USD 12 billion (approx. RM55.2 billion) in incremental sales, evidence that AI shopping agents are no longer experimental but part of large-scale retail operations.
Kate Spade’s AI Gift Concierge Signals Luxury Adoption
Kate Spade New York, owned by Tapestry, is among the first to use Amazon’s Agentic Shopping Assistant, launching an AI Gift Concierge focused on gift-buying. The conversational agent helps shoppers describe the occasion, style preferences and budget, then recommends products from Kate Spade’s catalogue in natural language. Amazon cites its data that 53% of shoppers report stress when buying gifts, so narrowing the use case to gifting tackles a specific pain point while keeping the experience on-brand and curated. Tapestry’s chief data and analytics officer Fabio Luzzi describes it as “a conversational AI experience that helps customers find the right gift, and it came directly from listening to our consumers.” The assistant was tested for around two and a half months before public launch, suggesting that enterprise deployments of AI shopping agents are moving into structured rollout cycles rather than one-off experiments.
Adidas Uses AI Agents to Power Ecommerce-as-a-Service
While Amazon is selling AI infrastructure through AWS, Adidas is building a parallel model on the brand side by offering ecommerce-as-a-service (EaaS) with AI agents from Salesforce. Senior project manager Dominik Seeberger explained that Adidas now operates full online stores for partners such as Audi’s Formula 1 team, running the Audi F1 ecommerce site on its behalf. AI agents help Adidas scale service, merchandising and operations for these partner stores without adding headcount, supporting 24/7 customer service via email and chat, global checkout in around 100 currencies, and fulfilment to more than 200 destinations. The Audi webstore looks and feels like an Audi property, while Adidas handles the underlying ecommerce stack and operations, directly mirroring how AWS shopping service customers keep their brand front-end while relying on agentic AI behind the scenes.
Why Agentic AI Is Reshaping Commerce Infrastructure
Taken together, Amazon’s AWS shopping service and Adidas’ EaaS model show a shift from single-site ecommerce to agent-driven commerce infrastructure. Retail AI assistants can now be deployed as modular services: Amazon sells the underlying AI shopping agents and tooling, while brands like Adidas operate complete ecommerce businesses for partners, powered by AI agents from providers such as Salesforce. This agentic layer spans both customer-facing journeys and back-end operations, from conversational product discovery and guided gifting to service workflows and global fulfilment. As more retailers pilot these systems, differentiation will come from how well AI is tailored to each brand’s catalogue, tone and service model, rather than from building raw technology. The next competitive frontier will likely be how smoothly these agents connect shopping, service and marketing into one continuous AI-supported customer experience.
Beyond Shopping: AI Campaign Agents and Retail Media
Amazon is also extending agentic AI into advertising via its Amazon Ads unit, which has released an AI Campaign Agent to help brands plan and manage ad campaigns. While details are still emerging, the move complements the Agentic Shopping Assistant by giving marketers AI tools to design, optimise and report on campaigns across Amazon’s media surfaces. The pattern is clear: the same core generative and agent technologies that power conversational shopping are being reused for campaign management, closing the loop between product discovery, purchase and advertising. For retailers, that means AI shopping agents and campaign agents can share signals about what customers ask for, what converts and which messages drive sales. Over time, this could turn retail media and ecommerce AI technology into one connected stack, with agents coordinating merchandising, experiences and media spend in near real time.
