Giving computers the capability to “understand” natural language was a prerequisite to make them real contributors to system engineering activities. Indeed, during the early engineering phases, concepts remains fuzzy, polysemic, and requirements are generally informal. Today, thanks to Large Language Models, this condition seems to be fulfilled and we can envisage teams of collaborating humans and AI agents. In this paper, we present our workflow and its supporting tools towards that goal. We describe and illustrate our approach on the functional analysis activities of a small robot.
Conejo et al. (Tue,) studied this question.