Legal discovery means that, when a lawsuit happens, both sides ask for (and receive) huge volumes of documents from the other party. Producing those huge volumes of documents can be quite challenging.<p>For example, someone in litigation with Google might hit Google with a discovery request saying "We want all documents, memos, and emails pertaining to search spam." That is, conservatively, millions of documents. So Google has to grab every email a Googler has ever written, <i>then</i> winnow that set down to just the relevant ones, then have lawyers pick through and exclude the ones protected by privilege, then get those hundreds of thousands of pages in a lawyer-accessible format delivered to the opposing party within a very tight timeframe.<p>This is challenging on Big Freaking Enterprise scales. Logik does the painful bits of it for you. They presumably charge Big Freaking Enterprise prices for it.
I don't know about logik.com specifically, but 2 critical problems in discovery are summarizing network graphs, and named entity extraction. Network graphs are, e.g. you take the Enron email dumps and you query for who emailed whom on what days, what people use statNet, SenSage and vertica for. Named entity extraction is distilling all the noun andpronoun phrases that could refer to one person or business entity.<p><a href="http://csde.washington.edu/statnet/" rel="nofollow">http://csde.washington.edu/statnet/</a><p><a href="http://37signals.com/svn/posts/2385-bootstrapped-profitable-proud-logik" rel="nofollow">http://37signals.com/svn/posts/2385-bootstrapped-profitable-...</a><p><a href="http://news.ycombinator.com/item?id=1414211" rel="nofollow">http://news.ycombinator.com/item?id=1414211</a>
It works by taking (mostly) off the shelf indexing software and attaching it to former executives of enterprise search and e-discovery companies.<p>You could reproduce the whole thing from scratch and not get a dollar's worth of sales because you don't have the relationships to get past the market gate keepers.