Basically, Silobreaker provides search results (particularly optimized for newsmakers and current-event topics) with a wealth of added-value elements including "context extraction" (how the person or topic fits in with other people, institutions, or categories), geographical mapping, trend tracking (graphing numbers of mentions in the world press), and relational mapping. The engine draws content on global issues, science, technology, and business from approximately 10,000 news, blog, research, and multimedia sources.
Easily the coolest part is the relational mapping, which displays in graphic form the web of people and topics in close proximity to the search term. By manipulating the various nodes in the map, you can see how the various points of interest relate to each other; by clicking on the midpoint between two nodes you can view an article or mention that explains how the two are related. (Here's a relational map Silobreaker built for me on Libyan Abu Laith al Libi, the alleged al Qaeda high-up killed this week in North Waziristan, Pakistan.)
The usefulness of this tool for a journalist like me is quite obvious. I think it has even more powerful uses in the corporate and government world, though, and not just for Web searches but for making and analyzing connections between data in various databases. Just think, for instance, how history might be different if the FBI had been able to use this tool to trace the connections between the known al Qaeda operatives who were attending pilot school in the United States in the late 1990s.