The rapid diffusion of automation and the application of artificial intelligence (AI) to production have boosted productivity and economic growth while also posing significant challenges for labor markets. These technological transformations reshape the tasks performed within occupations and, consequently, the skills that employers demand from workers, with potentially disparate effects across groups of workers depending on their exposure to automation and their ability to adapt to new skill requirements. Despite the growing international literature, empirical evidence for Latin America remains scarce. Given the region’s structural specificities, directly extrapolating results from other contexts may be misleading. This paper seeks to help close this gap by providing a robust, evidence-based understanding of the demand for soft, AI-related and non-AI digital skills in four Latin American countries, Argentina, Brazil, Colombia and Mexico, drawing on a novel regional taxonomy and more than one million online job postings. The results show that employers overwhelmingly demand multidimensional and highly complementary skill profiles, with technical non-AI and soft skills appearing jointly, in most cases, in the vast majority of postings. In contrast, explicit demand for AI-related skills remains limited, accounting for less than 3% of postings in all four countries.