Conversational AI for real-time life cycle assessment

Life cycle assessment (LCA) remains underused in early-stage design due to slow computation, complex interfaces, and reliance on expert interpretation. This study presents an AI-augmented LCA framework that couples a lightweight real-time computational engine with a conversational large language model to support rapid, standards-aligned exploration. Implemented in a modular Python–Flask architecture with CSV-based inventories, the system generates EN 15804–compliant A1–A5 results while maintaining sub-second recalculation. Users can conduct interactive scenario comparisons and one-factor-at-a-time sensitivity tests to probe the influence of material, transport, and energy assumptions. The conversational interface translates user queries into model operations and returns explanations using a complexity-adjustment mechanism intended to support both novice and expert users. In a pilot usability study (n = 14), the platform achieved an overall mean System Usability Scale score of 65.18, while experienced participants rated usability highly (SUS ≈ 92.5), suggesting additional onboarding and interpretive improvement are needed for novice users. Overall, the results establish computational fidelity and real-time exploratory capability within defined system boundaries, indicating the platform's potential for early-stage decision support while motivating broader benchmarking against rapid-LCA tools and larger, longitudinal user validation.