Thanks for sharing
I watched a bit of the video demo. It looks to me like if you want to automate a series of GUI tasks, you write a script that contains functions (like: switchApp, click, wait, and type) that take arguments that can be text or images.
Sample script from first video:
switchApp("System Preferences.app")
click(image of network system preference)
click(image of ethernet option)
click(image of "Using DHCP")
click(image of "Manually" from combo box / drop down)
wait(image of text field labeled "IP Address")
type("192.168.0.1\t")
type("255.255.255.0\t")
type("192,168.0.254")
click(image of "Apply" button)
Note the use of "backslash t" to tab between fields.
Automation via *Screenshots* - wow (based on jython)I guess there's a fair bit more to it than this -- there are tutorials and examples at:
http://sikuli.csail..../documentation.shtml 
