The digital transformation of archives, driven by artificial intelligence and machine learning, represents a fundamental methodological shift for historical research and archival practice. This paper examines two complementary technological approaches through concrete projects at UTwente, the Belgian State Archives, VUB, and UCLouvain. First, it analyses Automatic Text Recognition platforms - particularly Transkribus - which convert vast holdings of historical manuscripts into machine-readable, searchable data. Second, it introduces AI-rchivist, a prototype tool developed within the ARKEY project, which uses generative AI to extract metadata, generate multilingual summaries, and identify named entities.