Artificial Intelligence (AI) is a technology area creating high expectations in many enterprises when it comes to potential benefits for improving business, automating tasks or facilitating new kinds of services. However, the organizational take-up of "hyped technologies", such as AI, is accompanied by the danger that expectations regarding the benefits of the technology are not realistic which can result in failed innovation projects. This paper investigates expectations of two groups of enterprises from different application domains towards the organizational AI use. The results of the study indicate that often the technology is put into the focus rather than the organizational context which leads to neglecting requirements regarding process integration or data availability. As a consequence, the paper proposes a method component adding to enterprise architecture management, which describes how the organizational context of AI can be determined as support for AI requirements analysis and feasibility studies. The method component is demonstrated in a project aiming at introducing AI into instant payment transactions.