Retrieval-Augmented Generative AI in Technical Field Work: Knowledge Access, Work Design, and Human Judgment

Generative AI is increasingly used in organizational work, yet empirical research remains concentrated in office-based and digitally mediated settings. This paper examines a mobile retrieval-augmented generative AI system used by electrical grid field technicians to access internal documentation. Based on ten semi-structured interviews, we analyze how the system shapes knowledge access, decision preparation, and perceived work design. The findings show use mainly in uncertain, non-routine situations involving unfamiliar problems or difficult-to-locate documentation. The system supports faster access to organizational knowledge, helps technicians assess possible action, and enables bounded self-sufficiency, while requiring evaluation of output relevance, plausibility, and contextual applicability. The study contributes to research on AI and the future of work by showing how RAG-based support changes the knowledge conditions of technical field work and by specifying evaluative competence as the domain-situated capability through which workers assess and contextualize AI-mediated information before incorporating it into situated work.