Parallelism is always going to be required to support the computational demands of some problem domains, and will continue to be exploited via ‘traditional’ parallel processing methods and languages. However, this paper argues that parallelism will also become a requirement for the non-specialist user, but that the traditional parallelism languages and techniques do not have the right support for the engineering of large-scale, parallel applications. The paper discusses this issue, and presents a visual, object-oriented parallel programming language, Vorlon, which addresses the management of both problem domain complexity and implementation complexity, to support the development of general-purpose parallel applications by programmers who are non-specialists in parallelism.