[cte]

I realize now that we didn't show a stateful app, but basically an application is run in steps, and the boundaries between steps are the calls for user input because the app must pause to wait for input. So we store the last instruction executed and halt the app. Then when input is provided, the app is restored at the proper point and continues running. It is not ruby specific, but it is _much_ easier to build this with interpreted languages.

I don't think we have the killer app yet, but we implemented a mobile interface to google calendar with ~5 lines of code which I've been using for a little while, and its surprisingly useful. When all is said and done, I think we might want to simply pitch this as a platform for adding mobile features to your existing app, so it becomes more of a developer tool than a consumer facing product.

As for monitoring, yes, it becomes a tricky task to achieve high availability (and scalability), but its a really fun problem to tackle :)