Communication between the classical computing library and quantum devices is coordinated in PennyLane via a suite of plugins.

The main job of PennyLane is to manage the evaluation of parametrized quantum circuits (so-called variational circuits) on quantum devices, and to make them accessible to the machine learning libraries. PennyLane also provides access to gradients of quantum circuits, which the machine learning library can use to perform backpropagation, including through quantum circuits—an essential process for optimization and machine learning.