Which is cool. Because the "Semantic Web" has a chicken and egg problem - the standards, knowledge, &c... all exist; but no one is producing "semantic" content. Primarily because it's a pain in the ass to do so. Some people have a tried a mix of NLP processing with automated annotation of recognized keywords, triples, &c... but it never seems to catch on. So it really looks like Silk is trying to produce an attractive and easy to use semantic editor.
Once you have people producing semantic content, then other services like semantic search can make use of it - Silk also appears to be providing analytics and analysis software for the content you do create.
What I don't like about this is (I'm guessing here) it remains on their servers; so instead of using Google Docs you would use Silk. I guess.