What comes next might be a phase of lots of experimentation made possible by open source unstructured databases like CouchDB.

CouchDB is an up-and-coming schema-less document database that just ended its incubation status and reached top-level project status in the Apache organization. The API is HTTP-based, the default GUI is HTTP-based (though you can proxy it), and information is serialized in the JSON format. Because CouchDB applies higher order functions in the map-reduce pattern incrementally on the data, it has the potential to handle search and retrieval of large data sets at a logarithmic rate of performance.

If the UI is just data stored with in the same way as the data manipulated by the UI, you might get a paradigmatic example like this: http://jchris.mfdz.com/posts/128

-Sam