On May 27, Amazon Web Services (AWS) announced the launch of the “AWS Agentic Shopping Assistant” (ASA). For the first time, AWS is making available to external retailers the underlying AI shopping architecture, code, and practical know-how that powers Amazon’s own Alexa for Shopping, allowing them to customize and deploy it based on their brands and product catalogs. AWS claims that implementation can be completed in as little as 60 days. The first client to adopt ASA is Kate Spade, a fashion brand under Tapestry; in April this year, the brand launched an AI gift advisor powered by ASA. This tool engages consumers in conversation to learn about occasions, recipients, and style preferences before generating tailored recommendations. Under the hood, it utilizes Anthropic’s Claude Haiku 4.5 model hosted on Amazon Bedrock, and after roughly two-and-a-half months of testing, it officially went live. Amazon also noted that numerous other retailers are currently in the testing phase. Last year, Amazon’s AI shopping tools generated approximately $12 billion in additional sales, and in 2025 over 300 million users made use of them.
This move aligns with Amazon’s long-standing strategy of commercializing its internal technologies; previously, AWS rolled out cloud computing, cashierless checkout systems, and logistics services using similar methods. However, potential competitive concerns remain: retailers must entrust their AI shopping infrastructure to Amazon’s cloud division, which itself competes directly with those very brands in the retail space. AWS emphasizes that retailers utilizing ASA retain full control over customer data, product catalogs, and business rules. Meanwhile, companies such as OpenAI, Google, and Perplexity are also developing AI shopping tools, though some initiatives have faced setbacks due to technical hurdles or high barriers to entry for retailers.