Uber Cart Assistant

Cart Assistant turns a shopping request or uploaded image into a draft grocery cart that the shopper can review and edit. Uber Engineering explains that the system divides the job between language models and conventional software: models interpret the request and judge suitable products, while inventory services and rule-based checks handle availability, prices and valid cart construction. The engineering account describes this as a sequence of focused steps.

Quantity selection shows why that division matters. A request for twelve eggs needs to become an appropriate purchasable package. The model interprets packaging and quantity context; arithmetic and validation check the resulting units. The workflow ends with the shopper's decision to place the order.