At the core of the French royal memoria since the Middle Ages, the Trésor des chartes are a perfect case study to identify the potentials and challenges both of and for using traditional regestes and artificial intelligence. Our paper will highlight our work with the manuscript registers belonging to the Trésor des Chartes and the analytical inventory thereof. Since its inception in 1958 under the supervision of Robert Fawtier, it has gained recognition for its exceptional quality and remarkable precision in handling the contents of these volumes (Paris, National Archives, JJ series). Despite these accomplishments, the inventory remains largely incomplete, as the printed inventories solely encompasses registers JJ 37 to JJ 79B, followed by unpublished, manuscript inventories until JJ98. Additionally, only the first and third printed volumes feature an index facilitating the exploration of subjects, place and person names. To address these limitations, researchers typically resort to supplementary materials such as archival cards, selective inventories and editions, and occasionally resort to perusing the registers themselves extensively.