Abstract Architectural decision‐making often requires navigating a wide array of disparate evidence sources, from heuristic lists to empirical studies, to anticipate how choices will impact the exhibition of future ilities. This paper presents a proof‐of‐concept utilizing Large Language Models (LLMs) as a cognitive assistant to assist engineers in synthesizing architectural evidence efficiently. By automating evidence retrieval and analysis from diverse data pools, the proposed approach enables a more informed, streamlined decision‐making process for system architects, much like how TurboTax aids individuals in generating accurate tax returns. This novel approach aims to improve accessibility to architectural insights, bridging the gap between human expertise and the vast reservoirs of knowledge available at organizations, ultimately enhancing the effectiveness of architectural choices in complex systems.
Salado et al. (Tue,) studied this question.