Introduction: The rapid adoption of generative artificial intelligence (GenAI) is transforming organisational work practices and creating new demands for capability development. Despite increasing investments in AI, organisations often struggle to train employees sustainably. Training initiatives are often fragmented, inadequately adapted to varying levels of employee readiness, and weakly integrated into broader organizational structures, limiting both learning outcomes and long-term adoption.

Research Question: This study addresses the research question: “What requirements should inform the design of a systematic and sustainable method for training employees in generative AI tools within organisations?”

Method: The research follows a Design Science Research (DSR) approach and focuses on the requirement elicitation stage as a foundation for future artefact development. Data were collected through a structured review of scientific literature and eleven semi-structured interviews with employees, managers, and domain experts. The empirical material was analysed using thematic analysis to identify recurring challenges, needs, and expectations related to organisational AI training.

Results: The findings reveal recurring deficiencies in training practices, including inadequate readiness assessment, limited opportunities for handson practice, weak contextual relevance, discontinuous learning processes, and insufficient organisational support. To address these, five central requirement dimensions were identified: readiness diagnostics, differentiated learning pathways, experiential and hands-on learning, task-specific contextualisation, and organisational scaffolding. These dimensions were translated into a structured set of requirements and represented through a goal model.

Discussion: Effective GenAI training cannot depend solely on access to AI tools or isolated instructional sessions. Instead, training must be embedded within broader organisational learning structures that align individual and organisational learning needs. By providing a theoretically grounded and empirically informed set of requirements, this study establishes a foundation for the future development and evaluation of methods aimed at supporting sustainable AI capability development within organisations.