83: Bioinspired Modeling Using Llm-Based Agents

Recently, the Digital Materialities (MAD) course at the Faculty of Architecture, Design, and Urbanism (UdelaR, Uruguay) has focused on nature as a primary source for design inspiration, exploring the geometric logic inherent in biological forms. Its latest edition implemented a systematic geometric analysis of natural structures to identify underlying patterns and translate them into innovative design processes. To achieve this, the course employed advanced AI tools, specifically multimodal Large Language Models (LLMs), to analyze and generate geometric patterns from visual and textual data. This approach bridges the empirical observation of nature with its translation into rule-based computational models. These rules become fundamental inputs for developing generative algorithms, exploring how AI can enhance the creation of models that emulate the complexity and efficiency of natural systems. This work moves toward conceptualizing LLM-based agents within a design context.