AWS Machine Learning describes a VLM-based quality system for Amazon A+ Content

In an AWS Machine Learning blog post, the company described a pipeline for evaluating Amazon A+ Content by rendering each product page as a single snapshot and scoring it with a vision-language model. Those semantic scores are combined with OCR readability signals in a CatBoost classifier trained on expert annotations. SHAP then explains the decisions and produces actionable recommendations for selling partners. On 5,478 pages across 10 categories, the fused model reached 0.94 accuracy and 0.95 F1, and revised pages improved by 1.4 points on a 10-point scale.