Small parcel carriers are now basing shipping charges primarily on the dimensions of shipping cartons rather than the weight of packed cartons. Consequently, reducing wasted space in cartons can result in significant savings in shipping costs as well as corrugated material and dunnage costs. Most companies maintain a relatively small number of carton sizes, so the opportunity to select a carton to contain order items with a small amount of empty space is limited. Furthermore, choosing the best carton to pack items, either before or after they have been picked, is difficult for packers using the ‘eyeball’ method of selection and often requires repacking, which further adds to costs. Maintaining a large number of different carton sizes (eg 40+) increases the likelihood that a carton selected for packing will have minimal wasted space. This paper describes the challenges and artificial intelligence (AI)-based solutions developed to enable the determination of which carton sizes to maintain in a large set of different carton sizes, automatically calculate the best carton(s) for packing an order from the large set, quickly identify the best carton for a packer, efficiently replenish consumed cartons from a carton supply area, and simplify vendor managed inventory (VMI) by a carton supplier to enable frequent delivery of replenishment cartons.