From Chatbots to Agentic AI Shopping Assistants
An AI shopping assistant is a conversational system that can understand natural language, reason over product data, and independently manage multi-step ecommerce tasks such as discovery, comparison, decision support, and post-purchase help, creating a guided, human-like shopping experience across the entire customer journey. Retailers are moving beyond scripted chatbots toward agentic retail AI agents that can act on their own. Instead of only answering simple FAQs, these systems search catalogs, refine results, orchestrate virtual try-on technology, and support checkout decisions in one flow. This shift is visible in how shoppers now start with open-ended prompts like “I need a gift for a friend who loves minimalist home décor” rather than typing narrow keywords. For retailers, the promise is higher conversion and lower service load as AI takes on repetitive work, while still feeling like a personalised digital sales associate.
Kmart’s Joy: Virtual Try-On Meets Conversational Commerce
Kmart’s new Joy AI shopping assistant shows how agentic AI can reshape product discovery and decision-making online. Available on both the website and app, Joy combines conversational search with visual tools, including virtual try-on technology for selected products and a “See It in My Space” feature that places items and furniture into a customer’s own room before they buy. Shoppers can ask for specific sizes, styles, colours, and budgets, then see options within seconds, plus side-by-side suggestions for easier comparison. They can also upload photos to receive tailored recommendations, making ecommerce personalization feel more like an in-store styling session. Joy works across a wide marketplace range, covering Kmart, Target, and other brands, and is powered by Google Cloud’s Gemini Enterprise for Customer Experience, an agentic solution designed to manage AI agents from discovery through post-purchase resolution.
Amazon’s Agentic Shopping Assistant and the Kate Spade Gift Concierge
Amazon is exporting the technology behind its own AI shopping agent to other retailers through the Agentic Shopping Assistant on AWS. The solution, built on Amazon Bedrock, AgentCore, and OpenSearch, gives brands a ready-made technical base to launch retail AI agents “in weeks” instead of the years needed to build from scratch. Kate Spade New York is among the first adopters, using the platform to create an AI Gift Concierge that holds natural-language conversations and recommends gifts by occasion, style, and more. Powered by Anthropic’s Haiku 4.5 model and informed by questions previously asked to Alexa for Shopping, the concierge aims to feel “less like search and more like talking to someone who knows the brand and knows how to give a great gift.” Amazon says over 300 million customers used its own AI shopping assistant last year, validating the underlying interaction patterns.
What Makes Retail AI Agents Agentic, Not Just Chatty
The current wave of retail AI agents is defined less by chat and more by autonomy. Agentic systems can interpret open-ended intent, call multiple internal tools, and complete multi-step flows with minimal hand-holding. In practice, this might mean asking for “a small sectional that fits a narrow living room, under a certain budget, in neutral tones,” then seeing a curated set of sofas virtually placed in your home, along with alternatives and accessories. On Amazon’s side, AgentCore acts as the operational layer that lets agents search, retrieve, reason, and respond consistently at scale. On Kmart’s side, Gemini Enterprise for Customer Experience coordinates Joy across discovery, product comparisons, and resolution. The result is retail AI agents that behave more like digital store associates than FAQ bots, able to guide shoppers from vague needs to confident purchase decisions in one continuous conversation.
Implications for Checkout, Conversion, and Customer Experience
As AI shopping assistants become embedded across the funnel, the checkout experience is starting to look less like a static form and more like a guided dialog. Instead of bouncing between filters, size charts, and reviews, shoppers stay in a single conversational thread where the agent clarifies needs, suggests alternatives when items are out of stock, and reduces uncertainty with virtual try-on technology or in-room previews. According to Amazon, Agentic Shopping Assistant “was validated using billions of real shopping interactions on Amazon.com,” giving retailers access to patterns that have already driven successful outcomes. For brands like Kmart and Kate Spade, this move signals a broader transition from basic chatbots to autonomous retail intelligence systems. The competitive edge will come from combining shared AI foundations with each retailer’s own catalog, tone of voice, and customer insight to create distinct, loyalty-building experiences.
