AWS for Industries

Introducing seven new features of the Agentic Shopping Assistant on AWS

Earlier this year, we announced the Agentic Shopping Assistant on AWS, a generative AI-powered solution that helps retailers give online shoppers the kind of expert guidance they would get from a knowledgeable in-store associate. The Agentic Shopping Assistant on AWS brings the expertise and insights behind Amazon’s successful Alexa for Shopping AI assistant to retail customers. Shoppers describe what they want in natural language, and the assistant responds with personalized recommendations as rich product cards, complete with images, pricing, and quick replies. It answers questions about shipping, returns, and store policies, remembers preferences across visits, and speaks in each retailer’s brand voice. The result is a guided experience that lifts conversion, reduces returns, and deepens the retailer’s direct relationship with its customers, rather than losing product discovery to general-purpose answer engines.

Retailers choose this solution because of what it gives them beyond the technology. They own the code and can innovate according to their unique needs, with full transparency into how the assistant works rather than a black box. Customer data is never shared, and everything runs in the retailer’s own AWS account, while shopper intent signals flow back unfiltered to improve merchandising and marketing decisions. And because retailers pay for the underlying AWS services only, the cost is often lower than commercial off-the-shelf alternatives.

We are excited to share a set of new features we have shipped since our initial announcement of the Agentic Shopping Assistant on AWS. Since launch, retailers have started building with the solution and telling us what they need to go further, and we have been listening. What we build reflects two inputs: demand signals from AWS retail customers globally, and inspiration from Amazon’s continued innovation with Alexa for Shopping:

  • Self-service admin portal. Of all the features we have shipped, this one has drawn the strongest response from retailers. A dedicated portal now gives business users direct control over the configuration changes they make most often, without waiting on development cycles to deploy them. Teams can manage brand voice prompts, ingest catalog updates and standard operating procedures, plug in external Model Context Protocol (MCP) tools using a dedicated configuration tab, monitor tool health, and track every change for governance and auditability, with version history and one-click restore when a prompt drifts too far.
  • Long-term memory. The assistant now remembers prior conversations and stated preferences across sessions. By using Amazon Bedrock AgentCore memory and conversation history persisted in Amazon S3, returning shoppers pick up where they left off. Combined with the User Profile Awareness feature, which draws on customer data platform attributes such as loyalty tier, recent purchases, preferred categories, sizes, and typical price range, the assistant knows both who a shopper is and what they have talked about. Every returning visit starts warmer and more relevant, which shortens the path to purchase and makes the experience feel like a relationship rather than a fresh search each time.
  • Multimodal search and voice-to-voice. Shoppers can upload a product photo and the assistant finds visually similar items from the catalog, interpreting color, style, and category to return a curated carousel. They can also speak to the assistant instead of typing, and with voice-to-voice, powered by Amazon Nova Sonic, the assistant speaks back for a true conversational experience. Shoppers can search by image, voice, or text—whichever fits the moment—capturing intent that words alone often miss (especially in large, visually driven catalogs).
  • Conversation session history selector. Shoppers can now browse and revisit their past conversations from a chat history panel, organized by date with auto-generated topic labels, and pick up any earlier session where they left off. This gives shoppers control over their own history and lets them return to an earlier consideration set without starting over or repeating themselves, which supports the longer, multi-visit decisions common in retail purchases.
  • Cart awareness and management. Shoppers can now move from discovery all the way to purchase without leaving the conversation. They can view their cart anytime, by button or in natural language. Shoppers can add or remove items just by asking, and the assistant can suggest complementary items, flag duplicates, and offer budget alternatives. Advanced cart actions, including quantity updates, size and variant selection, round out the experience. On the retailer side, the assistant plugs into whatever cart infrastructure already exists and supports both logged-in and guest shoppers with no migration or rebuild required. The assistant now carries the shopper through checkout and creates a natural surface for relevant add-on suggestions.
  • Shopify integration. Retailers on Shopify can now connect their existing catalog to the assistant through the Universal Commerce Protocol (UCP) with no custom API mapping required. Shoppers get full product discovery and a streamlined cart experience, and Shopify-based retailers get to production faster by taking advantage of this simplified integration.
  • Conversational intelligence dashboards. A dashboard system gives retailers better business insights, with real-time demand signals, trending categories, catalog gaps, AI-generated insights, and session replay, alongside an operations view covering latency trends, token usage, tool error rates, and guardrail triggers. Retailers get visibility into both what shoppers want and how the system is performing, turning every conversation into a source of merchandising and operational insight.

Conclusion

Customer expectations are shifting toward conversational, AI-guided shopping, and the pace of that shift is why we keep building. The Agentic Shopping Assistant on AWS gives retailers a proven foundation, and pairs it with the domain expertise, catalog, and brand voice that only the retailer can bring. We will continue to innovate on behalf of our customers beyond these seven features, shipping new capabilities that help retailers deliver shopping experiences that feel personal, visual, and complete from the first question through checkout. To learn more about the Agentic Shopping Assistant on AWS, check out the solution listing or talk to your AWS account manager.

Resources

Vince Koh

Vince Koh

Vince Koh leads worldwide strategy and thought leadership for Digital Commerce at AWS. In partnership with the AWS Retail and CPG leadership teams, Vince works to shape and deliver go-to-market strategies and innovative partner solutions for consumer enterprises looking for guidance on how to transform their businesses with new capabilities for online, social, and mobile commerce, and how to connect the dots to create a unified commerce experience. Across his 15-year career, Vince has led digital commerce for both global enterprises and high-growth startups, developing and executing direct-to-consumer (DTC), marketplace, and omni-channel retail initiatives. Prior to joining AWS, Vince served as SVP of Commerce & Conversion at Weber Shandwick; VP of Ecommerce for Iconix Brand Group; led merchandising, strategy, and operations at venture capital backed startups (Keaton Row & Fab); and led global retail consulting projects at Accenture. He holds an MBA from Cornell University’s SC Johnson Graduate School of Management.

Abhijit Dutta

Abhijit Dutta

Abhijit is a Senior Solutions Architect at AWS, specializing in Retail industry. He partners with enterprise retailers on their digital transformation—modernizing legacy platforms, enabling data-driven decision-making, and deploying generative AI solutions. His recent work focuses on the evolving spectrum of conversational commerce, helping brands create AI-powered experiences that guide customers from discovery to purchase.

Andrew Ang

Andrew Ang

Andrew Ang is a Machine Learning Engineer in the Amazon Machine Learning Solutions Lab, where he helps customers from a diverse spectrum of industries identify and build AI/ML solutions to solve their most pressing business problems. Outside of work he enjoys watching travel & food vlogs.

Ramesh Venkataraman

Ramesh Venkataraman

Ramesh Venkataraman is a Senior Solutions Architect at AWS who enjoys working with customers to solve their technical challenges using AWS services. Outside of work, Ramesh enjoys following stack overflow questions and answers them in any way he can.

Suprakash Dutta

Suprakash Dutta

Suprakash is a Senior Solutions Architect at AWS. He focuses on digital transformation strategy, application modernization and migration, data analytics, and ML. He’s part of the AI/ML community at AWS and designs GenAI and intelligent document-processing solutions.