The ever-present pressure on shortening the development cycles in automotive industry requires further development of numerical methods and software tools in support of those development processes. From the perspective of system dynamics analyses and control system designs, the development of methods that allow for automated generation of computationally efficient mathematical models is of a large interest. In the field of automatic transmissions (AT), automated modeling would ultimately include generation of a full-order, minimum-realization mathematical model directly from an AT schematics. Furthermore, it is of interest to derive a method/tool for automated generation of a reduced-order model for an arbitrary clutch state, which would account for the energy-dependent variables related to locked clutches. Having in mind the above objectives, the paper proposes a method for automated generation of the full-order AT model from a bond graph model that directly reflects the AT structure (i.e. its schematic). The proposed numerical method has been implemented within the 20-sim and MATLAB software environments, where 20-sim is used to draw the bond graph and export it to a MATLAB script (or simulate it). A proposed method relies on a system identification approach that extracts the characteristic state-space model matrices from either a 20-sim-Matlab exported script or 20-sim-simulated bond graph model. The automated modeling method is demonstrated on an example of an advanced 10-speed AT. The paper also outlines a numerical method of transforming the full-order model into the reduced-order model for a user-specified clutch state, including the related method implementation within the Matlab environment. The possibility of deriving the reduced-order model in the 20-sim environment is discussed, as well.