When WolframAlpha launched, I (and many others) immediately started putting it through its paces. It’s only been a few days since the launch and already some things have changed. Here’s what I wrote for my "Dollar Sign" column in ONLINE (that will be published in the July/August 2009 issue of the magazine):

"A new search engine, WolframAlpha debuted mid-May 2009 to overwhelming excitement—so overwhelming that it couldn’t always keep up, delivering an error message that necessitated familiarity with the film 2001. "I’m sorry, Dave, I can’t do that." Dubbed a "computational knowledge engine" rather than a search engine, WolframAlpha attempts to provide answers to questions. Not any questions, though, but the ones that can actually be computed.
 
One of the suggested topics in "Money & Finance" and one of the suggested searches is to compare publicly traded companies by entering their stock symbols. Enter, say, BAC WFC (Bank of America and Wells Fargo) to retrieve some very interesting comparative statistics, including a mean variance optional portfolio, something you don’t see every day. Entering just one stock symbol retrieves more information than the comparative statistics, but if you’re doing a full company review, supplementing what WolframAlpha gives you with a company overview from Google Finance, Yahoo! Finance, Hoover’s, LexisNexis, and Factiva is definitely in order.
 
WolframAlpha has a promising future, but presently it needs a lot of work. How can a search on "oil gas industry "not recognize the word oil? And why would "home foreclosures" suggest "home disclosure" as the proper search strategy? Falling back to simply entering "foreclosure" gave me the definition of the word, its pronunciation, and broader terms, none of which was useful.
 
When WolframAlpha does give you an answer, it supplies a list of sources. Strangely, it stresses that those sources may or may not have been used to provide its answer. I’m intrigued by the notion of computational search, but concerned that it will not be robust enough to engage business researchers’ interest."
 
Of course, looking at WolframAlpha again today and reprising my "oil gas industry" search, I got a completely different answer. It now "interprets" my request as "energy" — that’s it, just one word! Following that is a list of companies (expandable to 238) that WolframAlpha thinks are in the energy industry. WolframAlpha still suggests "home disclosure" for "home foreclosures," though. Since this is a computational engine, I tried something that required a numerical answer: "GDP Czech". That gave a nice, tabular answer, in both dollars and koruna, plus it added in real growth and the unemployment rate. One problem: the data was from 2007. WolframAlpha relies on "curated sources" and my suspicion is that many of these are in the scientific/mathematics arena rather than business. It will be interesting to see how this "knowledge engine" develops.