This presentation explores the potential of artificial intelligence tools to automate certain recurring tasks in the archival field. It is based on concrete examples involving data processing in Excel or CSV files, directory organization, and metadata extraction. By formulating needs in natural language, it becomes possible to automatically generate appropriate formulas or scripts—for instance, to normalize dates, rename files, create folders, or extract relevant values from structured rows. The approach relies on a structured prompt method, designed as a translation of documentary rules into instructions interpretable by an automated agent. Beyond saving time, this method helps strengthen operational control, ensure traceability, and support progressive skill development. In this way, artificial intelligence becomes a tool for supervision, documentation, and professional learning.