Interaction nets are a type of graph rewrite system. Introduced initially as the basis for a new programming paradigm, they have attractive properties, including automatic memory management, direct manipulation of data structures, natural parallelism and opportunities for a visual programming style. However, their main contributions have been investigating computation models, most notably in the optimal reduction of lambda-calculus expressions. Interaction net programming languages have been worked on for over 30 years but remain low-level. We are working on a new approach, using functional programming techniques, to create a high-level language that is isomorphic to interaction nets and their rules.