In a traditional JVM implementation, a class-file will compile down to bytecode, and the bytecode interpreter turns the relevant ops into calls to the runtime routines.
JNI provides another way to call the runtime routines (using C++ calling conventions), and for the runtime to call back into your code.
So, for example, if you have a method that adds two numbers together defined on the Foo class, then you want to instantiate that class and call that method from your app, you would compile the "add two numbers" method down to machine code. You'd pass the location of the compiled method to the runtime, registering it with the Foo class, using JNI. Then you'd compile your app code (that will instantiate the class and call the method) to machine code with instructions to set up the stack and jump to the JNI routines to instantiate the class and finally call the method.
Make sense?
This most likely makes the same assumption that you are using a similar kind of project management as in RubyMotion for iOS and using RM specific libraries.
And when they start off with "Ruby classes are Java classes", then later say "everything is compiled to native" that seems like more double-speak.
But then later it says "The runtime uses the Java Native Interface (JNI) in order to integrate with Java".
And then later it says, "RubyMotion Android apps are fully compiled into optimized machine code, exactly like their iOS and OS X counterparts."
Does any of this make any sense? They seem to be contradicting themselves left and right on the same page.