InvEvolve: Evolving White-Box Inventory Policies via Large Language Models with Performance Guarantees

TL;DR

InvEvolve leverages large language models to evolve inventory policies with statistical safety guarantees, outperforming classical and deep learning methods in dynamic retail environments.

Contribution

We introduce InvEvolve, a novel framework that uses LLMs with confidence intervals to generate safe, effective inventory policies in non-stationary settings.

Findings

InvEvolve outperforms classical inventory policies.

It improves upon existing benchmarks in real-world data.

Provides statistical safety guarantees for policy deployment.

Abstract

We study how large language models can be used to evolve inventory policies in online, non-stationary environments. Our work is motivated by recent advances in LLM-based evolutionary search, such as AlphaEvolve, which demonstrates strong performance for static and highly structured problems such as mathematical discovery, but is not directly suited to online dynamic inventory settings. To this end, we propose InvEvolve, an end-to-end inventory policy evolution and inference framework grounded in confidence-interval-based certification. Built on a large language model trained via reinforcement learning, InvEvolve can process demand data together with additional numerical and textual features, and generates white-box inventory policies with statistical safety guarantees for deployment in future periods. We further introduce a unified theoretical model that connects training, inference, and deployment. This allows us to drive a lower bound on the probability that InvEvolve evolves a statistically safe and improved policy, and to characterize the multi-period performance gap relative to the oracle-safe benchmark. Tested on both synthetic data and real-world retail data, InvEvolve outperforms classical inventory policies and deep learning based methods. In canonical inventory settings, it evolves new policies that improve upon existing benchmarks.