While computational modeling of shape-memory alloys have become almost routine, the available models are often adapted from other phenomena, do not account for the microstructural details and consequently applicable to a narrow range of experiments to which they have been fit. We present a new constitutive framework for shape-memory alloys that implicitly but realistically accounts for the underlying microstructural details and their evolution. For example wires, tubes and sheet are heavily textured and this can easily be incorporated into the model. Further, it can handle nonproportional loading including complex tension-torsion experiments. Furthermore, the model is capable of reproducing a number of instabilities that have been experimentally observed. At the same time, the model is relatively easy to use and may be incorporated into commercial computational packages. The main idea behind the model is that it builds on concepts that are derived from the micromechanical analysis of shape-memory alloys. These include a texture-dependent set of recoverable strains and a defect-sensitive kinetic relation.