- Macy’s “Ask Macy’s” chatbot, built on Google’s Gemini platform, recorded 4.75 times higher revenue per visit during beta testing, one of the stronger specific figures disclosed by a major retailer this cycle.
- Instacart, Gopuff and Walmart are each taking distinct approaches to AI-driven basket growth: in-aisle deal surfacing, pre-built cart suggestions and recipe-based intent inference respectively.
- AWS claims conversational shopping sessions convert at 3.5 times the rate of keyword search, but the original source for that figure has not been independently verified, retailers evaluating the claim should press for methodology.
Macy’s “Ask Macy’s” chatbot recorded 4.75 times higher revenue per visit during beta testing, making it one of the more specific AI commerce claims a major retailer has put on the record. That figure arrives alongside a cluster of AI shopping announcements from Instacart, GopuffWalmart and Amazon Web Services each taking a different route toward the same goal: personalisation that drives basket size and cuts friction at checkout.
Instacart’s Caper Carts Bring Real-Time Deals In-Store
Instacart and Weis Markets launched AI-powered Caper Carts this week in select Pennsylvania stores, with further rollouts planned through 2026. The carts run on Instacart’s Connected Stores technology and give shoppers real-time spend tracking, personalised on-cart coupons and direct integration with Weis Rewards. Location-based deals can be clipped mid-trip, at the exact moment a customer is standing in the relevant aisle. The underlying data layer combines online and in-store signals, giving Weis a live picture of customer behaviour, shelf activity and store traffic. The commercial logic is straightforward: deals surfaced at the point of decision convert more reliably than those pushed via email or app notifications.
Gopuff’s “Go” Assistant Builds the Cart Before You Open It
Gopuff launched “Go,” a personal shopping assistant built into its quick-commerce app. Rather than waiting for a customer to browse, “Go” analyses past purchases and contextual cues to infer intent and construct a preliminary cart before the user has searched for anything. The goal is to surface relevant products shoppers might not have consciously sought out, increasing average basket size while reducing the decision-making load at checkout. Whether the approach meaningfully lifts retention in a category where switching costs are low is harder to verify from the announcement alone.
Macy’s “Ask Macy’s” Chatbot Targets Cart Abandonment
The 4.75 times revenue-per-visit figure from Macy’s beta is the headline number, but the mechanism behind it is worth examining. “Ask Macy’s” is built on Google‘s Gemini Enterprise for Customer Experience. It scans Macy’s full product catalogue in response to a query, then asks follow-up questions about preferences such as colour, cut and fabric to narrow recommendations. The iterative back-and-forth increases time on site and, according to Macy’s, reduces cart abandonment. A virtual try-on feature lets shoppers superimpose items onto an uploaded photo. Beta figures are rarely representative of steady-state performance, so whether the revenue uplift persists at full scale remains to be seen.
Walmart Moves From Recommendations to Intent Inference
Walmart CEO John Furner described this week how the retailer’s AI agents are moving beyond product recommendations toward intent inference. If a customer adds ground beef and mozzarella to a cart, the system infers a likely recipe and surfaces the remaining ingredients without being asked, increasing basket size without requiring a separate search. Walmart is also deploying AI internally: an agent called Code Puppy is available to employees across communications and merchandising to accelerate technology projects. The combination of internal and customer-facing deployments suggests Walmart is treating AI as operational infrastructure rather than a front-end feature.
AWS Offers Retailers a Configurable AI Shopping Layer
AWS cites a conversion rate 3.5 times higher for conversational shopping sessions compared with traditional keyword search, though the original source for this figure has not been independently verified.
Taken together, this week’s announcements reflect a practical commercial focus: reduce checkout friction, increase basket size, surface the right product at the right moment. The Macy’s beta number is the strongest specific claim in the group. The Gopuff and AWS conversion figures need named sources before they can be taken at face value. Retailers with large proprietary order histories now have the raw material to build differentiated AI layers, execution at scale is where these deployments will be won or lost. For more analysis on enterprise AI strategy, visit our Enterprise AI section.
