In this talk, we explore how artificial intelligence is being integrated into Yale’s digital infrastructure to manage its vast collections, which include over 20 million objects and 15 million works. These are connected to over 6 million people and groups. Central to this work is the "Name Parser" project, which utilizes AI models trained on large synthetic datasets to transform unstructured and inconsistent historical name data into a standardized, machine-readable format. This structured data then allows for the application of advanced heuristics to automatically merge duplicate records or flag them for professional review, significantly improving the accuracy of Yale’s LUX discovery platform. Beyond data organization, the talk explores the use of Vision-Language Models (VLMs) for scalable and generalizable Handwriting Text Recognition (HTR), alongside creative applications such as the digitization of shadow puppets