Recently a new metaphor for diagramming, the intelligent diagram has emerged. This new metaphor provides an excellent basis for diagram beautification since it automatically collects geometric constraints which capture the semantics of a diagram during diagram construction. By applying visual-language-specific layout rules which impose additional desired constraints on the diagram component placement, we can provide powerful semantics preserving diagram beautification. We have demonstrated the feasibility of this approach by developing a simple extension to the Penguins system which generates beautification rules from a grammatical specification of the visual language and tested the system with three example visual languages: Binary trees, state transition diagrams and mathematical equations.