Back to AI News
arxiv

G-CAST: Governed Continual Self-Training for E-Commerce Agents

The paper presents G-CAST, a governed continual self-training system for LLM-powered e-commerce shopping agents that must adapt as user demand drifts and the agent's tools, skills, and procedures evolve. G-CAST converts daily production evidence into reward-routed, entropy-prioritized training data and uses governed reward evolution, with checkpoint transitions controlled by cumulative evaluation to add new behaviors without losing accepted ones. Experiments on a real production e-commerce agent reportedly show better adaptation and preservation of prior behavior than a strong autonomous post-training baseline, with ablations supporting the data orchestration and reward evolution components. The system is deployed in Yuancai, a B2B procurement agent on Alibaba's 1688 platform.

AI Brief

It offers a concrete deployment-tested approach to continual post-training for production LLM agents facing drifting demand and evolving tooling.

arXiv:2611.17814v1 Announce Type: new Abstract: With the rapid advancement of Large Language Models (LLMs), an increasing number of users are turning to LLM-powered agents for conversational shopping in e-commerce. However, the user needs continue to change as new product categories, procurement scenarios, and feature requests emerge. At the same time, the product's execution harness-including its tools, skills, and operating procedures-continues to evolve to support increasingly sophisticated interactions. These changes create two practical challenges: 1) Demand drift alters the distribution, intent, and complexity of user requests, causing a model trained once to fall behind; 2) Harness evolution requires the model to use newly introduced capabilities correctly without losing behavior that users already rely on. Periodic manual retraining is too delayed and labor-intensive for this setting. To address these challenges, we present G-CAST, a governed continual self-training system that converts daily production evidence under evolving demand and capability contracts into reward-routed, entropy-prioritized training data and governed reward evolution. Governed checkpoint transition under cumulative evaluation controls each model update, enabling the agent to learn newly introduced behaviors without sacrificing previously accepted capabilities. Experiments on a real production e-commerce agent show that G-CAST achieves stronger adaptation to newly introduced demand while better preserving previously accepted behavior compared with a strong autonomous post-training baseline. Ablations further show that both reward-routed, entropy-prioritized data orchestration and governed reward evolution are important to these gains. G-CAST is now deployed in Yuancai(http://cai.1688.com/), a B2B procurement agent on Alibaba's 1688 platform.
Be the first to rate this article.

Comments & Questions (0)

Captcha
Please be respectful — let's keep the conversation friendly.

No comments yet

Be the first to comment!