This is a frequently asked question by people who are just getting into the prototyping or production of digitally created models. The common misconception is that a CNC milling machine can simply "swallow" a complete 2D or 3D model and automatically produce a part. This misconception is aided by the fact that there are now 3D printing processes that can do just about that, practically all you have to do is hit the "Print3D" button...
Unfortunately, this is NOT the case with CNC milling...
CNC milling is considerably more complex than 3D printing. You cannot send designs to a CNC mill. You need to send specific machine code that tells the machine exactly what steps to execute in what order to make your part. This code is (in most cases) alphanumeric (a text file) and can be read and understood by ordinary humans as well.
1) Your "designs" in Rhino (whether they be in 2D or 3D) are used by some CAM software (which can run inside Rhino or as an external application) to generate what are commonly called "toolpaths". This step needs your input to some degree (depends on what CAM software you're using and the complexity of the part) - generally you need to at least choose a tool or tools and a type of machining strategy to use. Some CAM are more automatic than others, but you do need to decide certain things in any case.
2) Once the toolpaths are generated, you can see them on the screen, have the option to edit them, simulate them, etc. until you are satisfied with the result.
3) When the result is considered correct on the screen, you then run the entire machining sequence through another piece of software called a "postprocessor". This software is generally running inside the CAM program and is a translator from the "generic" toolpath format to your specific machine/CNC control combination (such as a Bridgeport model XXX mill with a Fanuc model YYY control).
The postprocessor will be different for every machine/control combination, and there are literally thousands. CAM software manufacturers usually provide a set of standard "posts" that can be customized, as well as a way of creating your own from scratch. The correct postprocessor outputs a text file with exactly the code the machine needs to generate the part.
4) This code can also be edited by the user before sending to the machine if desired - it's just text. It can be reviewed in any text editor like notepad, but there are also specific NC code editors which have specialized functions that make editing easier. There are more or less sophisticated NC editors available, ranging from free to moderately expensive.
5) The final code is transmitted to the machine. In many cases this is via a direct serial connection from the computer to the machine. In other cases (more recent controls) the machine is actually on the network, and the file is transferred via ethernet. In the first case (more common), there is another small piece of software that manages computer-machine serial communication. For NC files, this is normally called DNC software (Direct Numerical Control). Depending on the sophistication DNC software can handle form one to hundreds of machines simultaneously.
If you only have one machine, although it's a bit more cumbersome, you can even use something like built-in Windows Hyperterminal to transmit files. There are also some free DNC programs out there, and DNC is sometimes included in an NC code editing program. If the machine can receive files via ethernet, DNC is not necessary.