Lost sales inventory control improved with language model design help
SabreAgent: Language Models at Design Time for Lost-Sales Inventory Control
Computational Engineering, Finance, and Science
Summary
Managing inventory to avoid lost sales is tricky because demand and delivery times can be unpredictable. The authors created SabreAgent, which uses a language model during setup to build special forecasting and stocking rules tailored to each product. Once ready, the system runs without calling the language model again, using those pre-built tools to decide how much to order. Tests on many inventory situations show that SabreAgent outperforms current best methods by a clear margin. The improvements mostly come from smart operational research techniques, with some gains from the language model’s seasonal insights and policy choices.
What this means in practice
- •For inventory managers: Use SabreAgent to improve order decisions for products with seasonal demand and uncertain lead times.
- •For supply chain planners: Apply SabreAgent’s pre-designed priors and policies to reduce lost sales and improve stock availability in multi-product settings.
Authors
Yang Liu, Yulin Huang, Xue Yu, Jiong Dong, Jianshen Zhang, Yongzhi Qi
Abstract
SabreAgent uses a language model at design time to construct two components for lost-sales inventory control: a product-specific seasonal prior and a validation-selected capped base-stock policy family. During operation, statistical forecasting and inventory optimization use these frozen artifacts to determine orders, with zero language-model calls. We evaluate the approach on the $1{,}320$ instances of InventoryBench. Under the benchmark's cost assumptions, the operations-research core draws on a zero-lead-time optimality result and a projected-inventory rule for positive deterministic lead times. The latter computes replenishment shortfalls by propagating inventory using sales along simulated demand paths. The seasonal prior adds forecast variants alongside the original forecaster, and the selected policy family handles stochastic lead times with order destruction. SabreAgent scores $0.6311$, compared with $0.5380$ for the strongest published baseline, and ranks first in all six benchmark cells. Ablations attribute most of the gain to the OR core. In the paired analysis, the seasonal component adds $1.79\%$ across the three real-data cells, and the search component adds $2.3\%$ across the two stochastic-lead-time cells. These results demonstrate how model-generated priors and policy structure can improve an OR controller through design-time use.