Tag Archives: Spanlink

What advice would you have given to a clerk in 1955 who hoped to make his or her mark by becoming proficient at preparing customer invoices? Or to a corporate middle manager circa 1975? How about to the aspiring travel agent of 1995? The list of jobs that have been pushed toward obsolescence by computers is a long one, though it’s not always easy other than in retrospect to spot whose neck is on the block. Since the list is still growing, the important question thus becomes: Who ought to be polishing up some new skills today? Look no further than your own management team. Computer systems are on the verge of making a quantum leap in brainpower. How much smarter can they get? Prepare to be surprised. Computers have transformed the business world–as well as everyday life–by being able to rip through vast oceans of text and numbers to pull out, manipulate, and transfer data in useful ways. But clever as they are, computers have never been able to truly understand the data they’re working with. Google finds the exact words you tell it to find, no more, no less. A database system can dig up customers in your records, but only according to the numbers and terms you specify. That’s beginning to change. New, powerful software is emerging that can extract something akin to meaning from data–and take action based on that meaning. Indeed, these systems are poised to perform the sort of knowledge-gathering, analysis, and decision-making chores that probably take up much of your day. What’s scary is that in just a few years, such systems will be able to analyze a lot more information than you can, as well as react much more quickly to a greater number of complex situations. What it boils down to is that computers are close to taking over a portion of what CEOs do now. And that will almost certainly change what human entrepreneurs and managers need to bring to the party. One company on the frontlines of these changes is iSpheres. Based in Oakland, Calif., iSpheres specializes in what’s known as “complex event processing” systems, which enable computers to spot an unfolding problem or opportunity–and act on it, automatically. The company’s software grew out of a 10-year Air Force-funded research project at the California Institute of Technology looking at how computers can help fighter pilots make better decisions based on an overwhelming flood of incoming data about their aircraft, enemy aircraft, anti-aircraft batteries, supporting forces, and so on. Business managers, of course, also need to react quickly and proactively to real-time data. “People think real-time decision making is only important in transaction-oriented environments like the stock market,” says iSpheres CEO Deepak Gupta. “But you might want to know when more than four customers haven’t had their calls returned. Or if the humidity inside one of your trucks is rising. If you wait until a problem has escalated, then you may already be dealing with actual losses instead of potential losses.” The iSphere system can respond to brewing problems or nascent opportunities by sending an e-mail or cell phone alert to the right employee–just as you, the CEO, might. It could even dash off a note to a customer, submit a restocking order, or redirect a shipment. The systems currently cost about $100,000, though Gupta expects prices to drop over time. Things will get really interesting when you add a new capability called “text mining” to the mix. Text-mining engines, which can cost as little as a few thousands dollars, take up where Google leaves off, searching articles, webpages, blogs, and e-mail (and eventually, even mobile phone calls or television broadcasts) for ideas and even emotions, rather than specific terms. “Most information today is not in any kind of structured database,” says Brett Shockley, CEO of Spanlink, a text-mining software company in Minneapolis. “Our technology can crawl all that information, parse sentences down to words, and compare the words with a lexicon of the English language.” The technology can help customer service agents and even customers themselves instantly find answers to questions buried in a sea of data. Health insurer Humana Military in Louisville, for example, places Spanlink’s search on its homepage. “HMO members usually have to spend an enormous amount of time refining their search and sorting through results to get to the document they’re looking for,” says John D. Jones, the company’s director of technology innovation. “Our search is restricted to providing five results, and 85% of the time the customer finds the answer there.” (I tested the Humana system with a vague question about reimbursement, and the first hit was right on the money. The same search on my HMO’s website yielded 1,034 hits, and I couldn’t find a clear answer among them.) Jones adds that the Spanlink-based search has cut down on service-agent training requirements and raised the quality and consistency of agent responses, and will eventually lead to a reduction in the number of service agents it employs. Meanwhile, a Watertown, Mass., company called Cymfony is developing an even more sophisticated text-mining application dubbed Orchestra. Cymfony’s software not only finds answers to users’ queries, such as “What do people think of blue soda?” but also points out patterns or trends within the results–for example, that people who live in big cities are more likely to want blue soda, or that people who want blue soda tend to hate pink soda. The software’s ability to highlight connections and contradictions in huge disparate sets of data can be critical to marketing, says the company’s CEO, Andrew Bernstein. “We can do market segmentation to show who the new users of a product are,” he says. Computers will be able to react based on the sum total of all available data–in other words, on the same information that informs a human being’s decisions today. How does this all add up to making your top executives obsolete? Things are really going to take off when complex event-processing systems like iSphere begin to merge with text mining and other monitoring applications. When that happens, computers will be able to react based not merely on what’s in your databases, but on the sum total of all available information–what customers are saying, what competitors are up to, what industry and popular trends are taking hold–in other words, on the same sort of information that informs a human being’s decisions today. Right now, the systems depend on managers to provide the “business logic”–that is, you have to “teach” the systems how to react to different data. But as more companies learn to do that, vendors will be able to sell systems with a certain amount of logic included. The next step will be enabling the systems to teach themselves the logic they need to make key business decisions, in much the same way managers react to a wide range of information. Spanlink, Cymfony, and iSpheres already include rudimentary “self-tuning” capabilities, and far more will be emerging from research labs over the next few years. None of this means that companies won’t need smart managers. It just means that managers will be relieved of much of the monitoring and troubleshooting that make up their days now. With this new technology in place, their emphasis will shift toward more strategic and creative decision-making efforts. Or it will, at least, until computers can start thinking strategically and creatively. Which, by the way, they are already learning to do. But that’s a story for another time. David H. Freedman, a Boston-based writer and Inc. contributing editor, is the author of several books about business and technology. (whatsnext@inc.com)