Usually the modelling starts with observing species occurrences and with finding environmental variables that are thought to influence habitat suitability. In any case, conceptual models help to find the main biotic and abiotic factors that control species distribution.
Species occurrence data can be obtained by complex field surveys at the one hand but can also be derived from existing database records. Environmental data is normally derived from digital maps, such as digital elevation models (DEM) or climate surfaces. These data often come from remote sensing or are also the result of a spatial model.
It also needs a model that links the species occurrences to the environmental variables. These models can be statistical, descriptive, logical, or rule based.
We also need tools to process and store the input and output data. There are several tools available. Most of the preprocessing and data management could be done in a common GIS system like QGIS or ArcGIS. Another interesting software environment, primary for statistical analysis and graphics, is R. This software can be used for building the models but also for repetitive GIS processing.
Last but not least we need data and criteria to evaluate the predictions.
This was only a brief overview and I will get more in depth for each of the above mentioned points.