This paper presents a new approach for managing RDF repositories without necessary knowledge or experiences with RDF standard. The main differences between proposed solution and existing ones are adaptability on various semantic repositories, internal memory model for manipulation with RDF graphs and utilization of the Java annotations for mapping between ontology and java classes. Persistence API (P-API) represents Java library offered features for separation of business logic from RDF and allows simple integration of the semantic applications with selected RDF repository. The developers don't need to work directly with RDF query languages; they work only with Java classes. The P-API can also help developers with manipulation within RDF graphs constructed from data stored in repository, as the proposed RDF Memory model fits perfectly with such purposes. This Memory model can be used for more complex and straightforward manipulation with ontological knowledge models. The P-API was designed and implemented as a supporting solution on the middleware layer of the collaborative system in order to make bilateral communication effective and faster. The exploitation within KP-Lab system is described in a section no. 3 as an illustrative example for possible integration approach.