Automation of product definition in those Product Lifecycle Management (PLM) systems where lifecycle applicable product model is in the centre of engineering activities is important because reproduction of former models and propagation of change effects in the product model using conventional modeling methods are impossible. Automation requires representation of the relevant knowledge in the product model. Knowledge is activated when changes of the model and environmental effects on it requires repeated definition of affected features and parameters. In the knowledge of increasingly advanced product definition automation methods in PLM, this paper reports a recent result in product function and behavior driven automation of product definition. After an introduction to the state-of-the-art in product definition automation, a new methodology is explained for the application of deep knowledge and intelligent computing at higher level automation in product definition. This methodology is organically connected to existing methodology of PLM modeling systems considering recent level of product and knowledge feature definition in industrial PLM systems.