AWS launches agentic shopping assistant for retailers

AWS has launched an agentic shopping assistant for retailers, drawing on technology used in Amazon's own shopping tools.

The service combines architecture, starter code and advisory support through the AWS Generative AI Innovation Centre, allowing retailers to build conversational shopping assistants using their own catalogues, business rules and brand voice. AWS says deployments can be completed in about 60 days instead of taking years to build from scratch.

The underlying technology has already been used at scale across Amazon's consumer business. More than 300 million customers used the company's AI shopping assistant last year, generating nearly USD $12 billion in incremental sales, according to Amazon.

The launch gives outside retailers access to a system built on the same foundation as Alexa for Shopping. Each implementation is tailored to the client's own shopping environment rather than functioning as a generic chatbot, AWS says.

Retail push

The move comes as retailers look for ways to maintain direct relationships with shoppers as AI becomes a more common interface for product discovery and purchasing. AWS argues that branded assistants built on a retailer's own product and customer data offer an alternative to broader answer engines that sit between merchants and consumers.

AWS also points to stronger conversion in conversational commerce. According to the company, shopping sessions handled through conversation convert at 3.5 times the rate of traditional keyword search.

The pitch targets a wide range of merchants, from specialty retailers to restaurant chains and consumer goods brands. AWS argues these companies hold product knowledge and customer insight that broader AI platforms cannot easily replicate.

Kate Spade use

One early customer is Kate Spade, which used the AWS service to build an AI Gift Concierge. Developed by Tapestry, it is described as the first production-ready retail AI assistant built with Amazon Bedrock AgentCore.

The tool is designed for gift buying, an area where retailers often struggle to turn vague shopper intent into relevant product suggestions. AWS says the assistant engages users in natural dialogue about the occasion, recipient and preferred style before producing recommendations.

The project moved from contract to production in about two and a half months, followed by roughly the same period of testing before being made available to customers. AWS says the experience drew on shopper behaviour and questions asked by Amazon customers using Alexa for Shopping.

According to AWS, the Kate Spade system runs on Anthropic's Haiku 4.5 model and also uses Amazon Bedrock for observability, authentication and evaluations.

Yang Lu, Chief Information and Digital Officer at Tapestry, described the collaboration in a statement released at launch. "We are excited about the possibilities agentic commerce can bring to our customers. AWS brought the recipe, but together we built the customization our consumers needed," Lu said.

Amazon as test case

AWS says Amazon serves as "Customer Zero" for the components behind the new retail offering, meaning the software and architecture were tested inside Amazon's retail operations before being offered to external customers. That experience helped shape which functions and tools mattered most in live shopping environments, according to the group.

The new service is built on AWS products including Amazon Bedrock, AgentCore and OpenSearch. Those elements have been validated through billions of shopping interactions on Amazon.com, AWS says.

For AWS, the launch also expands its strategy of turning internal Amazon systems into commercial services. The company has long used its parent group's scale in retail and logistics as a proving ground for software later sold to third-party customers.

In this case, the proposition is not only technical speed but also data control. Retailers can keep their data within their own AWS accounts while adapting the assistant to fit their catalogue, customer base and brand language, AWS says.

Additional retailers are testing the system, according to AWS, though it did not identify them. The early emphasis appears to be on merchants that want to introduce conversational shopping tools without handing customer interactions to third-party consumer AI platforms.

The launch underscores how large technology suppliers are moving quickly to sell AI systems tailored to specific industries rather than general-purpose chat tools. In retail, the contest is increasingly focused on who controls the customer conversation at the point of purchase.

What took Amazon years to develop internally can now be deployed for other retailers in roughly 60 days with guidance from the AWS Generative AI Innovation Centre team, according to the company.