The preparation of data in a valuable and usable way is crucial for data-driven applications. The paradigm of data products was therefore introduced as a domain-centric, shareable, and valuable unit. To create relevant data products, it is essential to incorporate domain expertise. Consequently, low-code approaches have gained increasing attention, as they facilitate modeling of development steps through a graphical editor. Conceptual frameworks such as LALO and BARENTS address data product development through low-code approaches, but to translate these concepts into practice, a concrete implementation strategy is required. Therefore, different components of BARENTS and LALO and possible existing approaches were examined and discussed. A limitation in the way heterogeneous sources and sinks are connected to the system was identified. Current approaches fail to adequately address the complexity of heterogeneous sources and sinks by not making the connection process sufficiently accessible for domain experts. To address this limitation, an adapter-based approach was introduced in this work, allowing data providers and data consumers to define connections, mappings, and metadata. This ensures that the available data is both understandable and usable. In addition, it enables the involvement of domain experts in the development process. To demonstrate its applicability, a prototype with example implementations and a user interface was developed.