GenAI Designer: A Designer-Centric Personalized Fashion Design Platform

Large Language Models (LLMs) and Artificial Intelligence Generated Content (AIGC) are increasingly applied to fashion design. General text-to-image models trained on public fashion datasets often fail to capture individual designers' stylistic preferences. To address this limitation, we introduce GenAI Designer, a designer-centered expert system that integrates Visual Language Models (VLMs) with LLMs and generative image models to support personalized and interactive fashion design. GenAI Designer automatically extracts stylistic priors from a designer's past works as textual features.Next, the features are incorporated into multi-rounds design workflows, accommodating diverse multimodal inputs such as textual descriptions, stylistic cues, model photographs, and reference images. The style adaptation and instruction-following ability of GenAI Designer is verified over 200 professional runway images from the advanced WGSN database. The VLM referee's evaluation results indicate that GenAI Designer improves stylistic similarity by more than 4\% compared to baseline approaches, highlighting its potential to bridge personalized design needs with scalable AI-driven fashion workflows.