This work is part of the Digital Earth project.
This study explores the capture and valorization of tacit knowledge within scientific organizations, using BRGM as a case study. Facing the strategic challenge of knowledge management, we examine the transformation of individual knowledge into a collective asset via a three-pronged methodological approach: theoretical framework, use of AI tools to identify and transcribe this knowledge, and proposal of a CBR architecture with an AI agent ("beregem") to solve problems in geosciences. This research contributes to the management of scientific tacit knowledge through AI-based solutions