I'd pull JSON from the server-side via an Ajax request, where the data would look something like this:
You'd then have a generic JSON array you could walk, where you could dynamically add/remove BoxComponents to/from the Window.
text: 'This is a picture of a monkey flinging poo...'
text: 'Pics of snakes on a plane'
text: 'An Ant Eater eating what Ant Eaters eat...'