Different domain-specific subsystems must be integrated during the development of interconnected smart products, associated services, and service infrastructure. This results in increased interdisciplinarity and challenges for knowledge management across multiple stakeholders. These challenges start at the requirements definition phase, where mechanical, electrical, software engineering, regulatory, and service design requirements must be analyzed and managed to create a base for the smart product to be developed. These requirements are often modeled using SysML models to facilitate a multidisciplinary, holistic system model. The requirements, however, are usually available as texts in natural language as documents managed in PLM-Systems, making the requirement extraction and modeling time-consuming and prone to errors. This paper introduces a concept for leveraging LLMs to integrate holistic system modeling in SysML with the document and knowledge management capabilities of PLM-Systems. The aim is to facilitate improved knowledge management, time savings, and error reduction during multidisciplinary product development environments across multiple stakeholders in value creation networks. The knowledge assistant concept will be validated considering the use case of a 3D printer.