The linguistic equation (LE) approach uses the same compact structures are used for models and control systems. Inverse, internal and predictive control can be combined with switching and fuzzy set systems. Measurement levels, interactions and composite local models are analysed in a gradually re1ned way in the data-based modelling. In the applications, specific models and indicators are selected and constructed from similar building blocks. Intelligent analysers produce informative indirect measurements and indices for the controller which operates like an agent-based solution where all the actions are available for activation when needed. All subsystems are presented as parametric systems which can be tuned for wide operating areas by using a balanced set of process situations. The control solution has been tested in three different applications which use the same blocks in a process specific way.