In order for Semantic Web applications to be successful a key component should be their ability to take advantage of rich content descriptions in meaningful ways. Reasoning consists a key part in this process and it consequently appears at the core of the Semantic Web architecture stack. From a practical point of view however, it is not always clear how applications may take advantage of the knowledge discovery capabilities that reasoning can offer to the Semantic Web. In this paper we present and survey current methods that can be used to integrate inference-based services with such applications. We argue that an important decision is to have reasoning tasks logically and physically distributed. To this end, we discuss relevant protocols and languages such as DIG and SPARQL and give an overview of our Knowledge Discovery Interface. Further, we describe the lessons-learned from remotely invoking reasoning services through the OWL API.