The Robot Shopkeeper

Once upon a time, people often shopped in small, neighborhood stores where proprietors knew them personally and could anticipate their needs and knew their interests and favorite products. In today's world of online shopping, call centers and impersonal supermarkets, that human touch is missing. But technology developed by a former Cambridge University researcher could help introduce old fashioned personal care into online shopping.

Mike Lynch, who has a doctorate in pattern recognition, began developing algorithms to help identify patterns more than a decade ago. In 1991, he saw their commercial potential and launched Neurodynamics, which sold the technology to government agencies as an intelligence-gathering and law-enforcement tool. Five years later, his spinoff company Autonomy offered the technology as a means to ferret data from texts. Now, NCorp, another, recent spinoff, is developing consumer-management software based on Lynch's technology. It gleans information from structured databases, like tables, and organizes it to suit the customer.

The basic technology is a form of Artificial Intelligence that "gives computers the ability to recognize patterns the way humans can," explains Nick Bidmead, NCorp chief executive. "The human brain is very good at deciphering patterns, but there are limits as to how much it can process." A computer, however, that thinks like a human can run nonstop and scan infinitely more data. It needs does not need coffee breaks and never has an off day because of personal problems.

One client, AutoTrader, a dealer of used cars in Britain, is incorporating the technology in its web, interactive-TV and WAP sites. When a person is buying a car, there are myriad features to consider: cost, mileage, styling, make, color, etc. And the importance of each factor differs greatly from buyer to buyer. Using Lynch's technology, a prospective customer can describe his or her dream car, and the software will return with a list of those models that most closely match the ideal. U.K. telecom Hutchinson 3G, which hopes to begin rolling out its 3G wireless service later this year, is using NCorp software to improve the search and personalization of its content. Because a mobile phone's screen is so small, providers must ensure that the information returned is very relevant, Bidmead notes.

The software is also good at spotting problems. It can be used, for instance, to zero in on banking and insurance fraud. At the application stage for mortgages, loans and insurance policies, there are subtle differences between real applicants and frauds, and the software can pick up on those patterns. It is also being used on offshore oil platforms, monitoring such things as temperature, torque and speed to figure out when to stop drilling. If a drill bit stays in too long, the well can collapse; but if it's removed too soon, it won't hit its maximum potential.

One pattern that Bidmead can see without the aid of software is that despite its many potential uses, customer relations management is the technology's biggest market. By offering technology to help predict consumer behavior, NCorp hopes to increase its revenues by adding that humanizing factor to online shopping.