Instacart is betting on AI shopping carts that suggest what you should buy as you shop (www.businessinsider.com)

Instacart announced it will roll out Cart Assistant, an agentic AI shopping assistant, on Caper smart carts and retailer websites — beginning with Sprouts — to make real-time purchase suggestions and help shoppers while they shop. The Caper carts (acquired by Instacart in 2021) already scan items, display running totals, surface coupons and accept payment; Cart Assistant adds conversational, goal-driven features like building a party shopping list within a budget, flagging unwanted ingredients mid-shop, or checking your whole basket for allergens such as gluten. For the AI/ML community this is notable because it’s a move from backend recommendation engines into in-store, interactive agents that act on shopper intent. Instacart says it can leverage a large proprietary dataset — roughly $1.5B in orders and a catalog of 17M unique items — to train models for personalized suggestions, out-of-stock detection, and richer product analytics. The rollout signals growing commercialization of agentic and recommender systems at the edge of retail, raising practical considerations around on-device latency, integration with inventory systems, personalization quality, and privacy as more conversational, action-oriented AI hits physical retail.