King Abdulaziz Foundation (Darah) launched an advanced artificial intelligence system capable of reading and classifying ancient Arabic manuscripts with accuracy exceeding 95%. This system is the result of a three-year collaboration between Darah, Saudi technology companies, and research centers specializing in Arabic language processing. The system employs deep learning models trained on over one million manuscript pages featuring diverse calligraphic styles including Naskh, Thuluth, Ruq'ah, Maghribi, and Farsi scripts. It can recognize text in deteriorated and faded manuscripts that are difficult to read with the naked eye. Beyond text recognition, the system offers advanced features including automatic subject classification, estimated dating of manuscripts, and identification of calligraphic style and scribal hand. It also enables semantic search across digitized manuscript content. This project aligns with Darah's digital transformation strategy and its commitment to leveraging modern technologies in service of national heritage. It is expected to accelerate cataloging and documentation processes and facilitate research access to thousands of uncatalogued manuscripts.