I am using Sikuli extensively for UI and web test automation and think the image based approach is the future for test automation. In my next project, I will give the Kantu browser a try, the ability to run many tests at once would be a big help for us.
Typically the maintenance of our selenium tests require upwards of triple the effort to create the page object model foe the application.
This looks like a solution that allows my team members with less programming skills to contribute to the Web automation effort.