What they have done, and there are probably others as well, is write one piece of software to do everything. This makes it easier to learn but maybe a little less flexible than the usual 3 step methods. Usually you use one software package to draw your designs (CAD), then another one to turn the drawings into G-code (CAM), and lastly a third program to actually run your machine (CNC). For example you might do your CAD in AutoCAD, do CAM with MasterCAM and then send it to a Haas CNC for the high end route. Or CAD in Inkscape, CAM with a Python script, and EMC2 for the CNC, for the total open source route. While the line between CAD and CAM is pretty blurry these days, with many programs doing both, you still have to usually send it to Mach 3 or some other program to actually run the machine. If you want to get up to speed quickly and make money, then time becomes very important. Not only time to learn the different software, but the time to process from idea to metal. The all in one approach can be very fast from concept to cut. A lot has to do with what you plan on making. For example if you are doing production runs, then good nesting software can save you metal. If you're doing one offs, that won't really matter. Also your work might dictate the kind of software you need to run. If you get customer supplied drawings in some obscure file format, you will need software that can deal with it. Another factor is if you plan on using your CNC for things besides plasma. Adding a router lets you do other things, but if you have a plasma only package, it might not have the tools to deal with routing. Or maybe adding another type of CNC, and having it run on the same type of control software, so you can jump right in, with a short learning curve.