Nowadays, decision-support systems (DSSs) are considered as one of the most crucial parts of many systems whose goal is to use a set of data and make the optimum decisions. To develop high-quality DSSs that can easily be maintained, an architecture-centric software development process can be followed. However, we observed that the literature lacks an architecture modeling language for the high-level specifications of DSS architectures that can be automatically analysed and transformed into an executable code. So, in this paper, we propose a novel architecture modeling language called DSSGen which offers a set of component types (i.e., problem, diagnosis and action) for the high-level specifications of DSS architectures. We developed a toolset for DSSGen, which enables for (i) the modeling of DSSGen architectures via a modeling editor, (ii) simulating the architectural models using Modelica, (iii) generating quality Java code from the architectural models, and (iv) visualising the code structure using UML class diagram. We evaluated DSSGen in two parts. First, we measured DSSGen’s impact on the development time performance via four different case-studies that have been performed by a group of practitioners. Then, we conducted a survey with the same group of practitioners to understand their perspectives on DSSGen.