Order picking in manufacturing warehouses is a labor intensive activity with critical implications to the well-being of staff and operational performance of companies. This study addresses the need for applying digital technologies that lead to enhancing a human-centered approach in order picking. It proposes the use of artificial intelligence (AI)-enabled vision systems to facilitate the generation and analysis of information about tasks in manufacturing warehouses. We present the results of a collaborative project between academic and industrial partners from a case in automotive manufacturing. This consists of the development of a pilot study in a laboratory environment and includes two findings. First, we show the steps of implementing an AI-enabled vision system in order picking. This findings is important for automatically generating and analyzing information of tasks in order picking such as setup, travel, search, and picking of parts, which directly affect staff performance. Second, we discuss the implications of this findings for manufacturing companies and its contribution a future in order picking with improved human-centricity.