Integrating a graph database like Neo4j with an existing relational database system can be done in a way that doesn't disrupt the entire system, by understanding its place and contribution to the overall system. The primary database remains the system of record, while Neo4j serves as a secondary data store used for specific use cases such as OLTP mode for real-time queries and batch mode for delayed data processing. By designing an application without assumptions about the data source, domain objects are represented using POJOs, and their corresponding DAOs handle persistence to and from Neo4j. The graph model ties back to the RDBMS via a primary key, allowing the application to be testable independent of the data source. For initial import, data can be transformed into domain objects and saved, with options including SQL queries or an API exposed by the primary system. Once imported, the application is loosely coupled, and the problem of consuming data from multiple sources in multiple formats is reduced to parsing. An event-based integration allows for real-time updates, while periodic imports serve as a simpler solution depending on the use case.