We combine deep learning and network analysis to investigate the evolving conceptualizations of ‘women’ in professional and social roles through Harvard Business Review (HBR) articles from 1945 to 2023. We fine-tune BERT models on each issue of HBR, using these to predict contextual word associations and to construct both directed and symmetric semantic networks for each time period. Our resulting longitudinal semantic networks offer a unique window onto the progression of women’s representation in the business world, capturing subtle shifts in language use and conceptual associations. Preliminary findings reveal significant changes in the network position of ‘women’ post-1972, coinciding with major legal reforms. After this juncture, we observe increased eigenvector centrality coupled with decreased local clustering, suggesting a broadening of associations and roles linked to women in business discourse. Our analysis further suggests that while the markedness of ‘women’ persists, it has evolved into a more sophisticated form, serving both as a reminder of lingering distinctions and as a tool for addressing ongoing inequities. Our research offers a data-driven narrative of progress, while highlighting areas where linguistic and conceptual barriers may persist, with implications for scholars in business, gender studies, and linguistics, as well as for practitioners addressing gender dynamics in professional environments.