As others have pointed out, it might seem like an idea to "obfuscate" the source code, but really this does nothing to protect your application. For one, any obfuscation trick, no matter how obscure, is usually easily undone. Secondly, since Rails depends heavily on the use of eval(), there will be large chunks of raw code in your application even if, theoretically, it could be compiled.
Keep in mind you still have a lot of latitude in how you deploy Software as a Service. You can give your client DB dumps, you can run off your client's DB platform, if they want their data "in house" and not at some untrusted facility.
If you mean, "can I ship customers a Rails app that they themselves run, either on their own servers or an appliance, while keeping them from seeing the source code?", then the answer is "no".
You should be aware that the same answer applies to Python, Perl, Tcl, Java, and C#. Bytecode is trivially decompiled back to readable source code.
I don't know what the state of the project is, but it looks like what you wanted.
This system looks even easier to deal with than a typical J2EE app; at least with J2EE, I have to fix up all the JAD artifacts.
Beware of systems that offer packaged solutions to hard problems, like the software protection problem.
It's better to just host the application yourself.