Learn Methodology: Applying Artificial Intelligence To Lessons Learned Management

lessons learned management remains a challenge in project-based organizations, where valuable knowledge generated during execution often remains tacit or is lost after project closure. This paper proposes the LEARN methodology (Log, Explore, Analyze, Resolve, Normalize) as an operational approach that integrates the SECI and SYLLK models to structure the capture, validation, and reuse of lessons learned in engineering projects. Operational findings are transformed into explicit and normalized knowledge using corporate digital tools, while Large Language Models (LLMs) are incorporated as guided query agents to support contextual retrieval at the initiation of new projects, without replacing human analysis. The methodology was implemented and evaluated through a factorial experimental design, showing a significant increase in lessons learned captured and a strengthened organizational knowledge-sharing culture, with projected reductions in rework.