Converting biological strategies into practical design principles during the Discover–Abstract phase of the Biomimicry Design Spiral (BSD) presents a considerable obstacle, particularly for designers lacking a biological background. This research introduces a Retrieval-Augmented Generation (RAG) framework that combines a specialized AskNature database of 2106 documents with a locally executed Llama 3.1 large language model (LLM) to fill this void. The innovation of this study lies in integrating the BDS with a stage-specific RAG–LLM framework. Unlike BioTRIZ or SAPPhIRE, which require specialized expertise, our approach provides designers with semantically precise and biologically grounded strategies that can be directly translated into practical design principles. A quasi-experimental study with 30 industrial design students assessed three setups—LLM-only, RAG-Small, and RAG-Large—throughout six biomimicry design stages. Performance was assessed via expert evaluations of text and design concept quality, along with a review of retrieval diversity. Findings indicate that RAG-Large consistently yielded superior text quality in stages with high cognitive demands. It also retrieved a more varied array of high-specificity biological ideas and facilitated more coherent incorporation of functional, aesthetic, and semantic aspects in design results. This framework diminishes cognitive burden, boosts the relevance and originality of inspirations, and provides a reproducible, stage-specific AI assistance model for closing the knowledge translation gap in biomimicry design, though its current validation is limited to a small sample and a single task domain.

Car styling is crucial for consumer acceptance and market success. Since vehicle manufacturers produce electric vehicles, they have faced the challenge of maintaining the typicality of their original products and presenting the innovation of new technologies. We propose a method that integrates artificial intelligence (AI)-generated images and image classification technology to help designers effectively balance between typicality and novelty. We collected 118 pictures of electric vehicles and 122 pictures of fuel vehicles in 2024 from the BMW official website. Focusing on seven key visual features of the vehicles, we used the Waikato environment for knowledge analysis (WEKA) to train an image classification model on the dataset through three separate training and testing sessions. First, we used the prompts that described typical BMW design to generate images of new BMW electric vehicles in Stable Diffusion. The images consisted of 21 front views, 20 side views, and 20 rear views. The accuracy of the model of front views trained with the pyramid histogram of oriented gradients filter (PHOG)-Filter and random forest classifier was 78.5%, and the test accuracy reached 95%. The accuracy of the model of rear views trained with BinaryPatternsPyramid-Filter and random forest classifier was 80.5%, and the test accuracy was 90%. However, the accuracy of the model of side views did not reach 70%. That implies the distinction between BMW fuel vehicles and its electric vehicles is mainly based on the front and rear views, rather than the side view. The results of this study showed that integrating image classification and AI-generated images can be used to examine the balance between product typicality and novelty, and the application of machine learning and AI tools to study car style.