Using a theory pioneered by Vladimir Vapnik, a Russian mathematician and former researcher at the laboratories of telecoms giant AT&T, KXEN says it has developed a predictive and descriptive data modelling engine that can develop data models more quickly and cheaply than anything else on the market today. "We're talking of building a model in minutes or hours rather than days or weeks," says the company's UK operations manager Grayson Amies.
Vapnik theory is far from new. It was developed in the 1960s as a result of research into pattern recognition and has since been put to a number of uses. In KXEN's case, the theory has been adapted to enable the company to take data "as it is" from source data warehouses and to create "a corridor through the data", says Amies.
The end result of this, says KXEN, is that end-users can build analytical models more quickly than with traditional techniques. This is important, says Haddad, one of the three founders of the company and now its CEO, because it can dramatically cut the costs associated with data analysis. It can cost thousands to employ a statistician to build a model, he says: "When you embed KXEN into an operation the process cost [of building a model] is zero."
Such claims, the company says, are already being substantiated by customers. According to Haddad, one US bank that has adopted its technology used to only be able to build 12 models a month due to the cost and time restraints. Now, he says, it builds as many as ten a day.
Haddad stresses that KXEN's technology can be applied to any data set. "Forget about data mining; this is a natural human process to extract information from data," he says. Potential applications of the company's technology include fraud management, forecasting and risk and return analyses.
Sales have grown quickly in the last year, albeit from a low base, rising from €391,000 in 2000 to between €1.1 million and €1.7 million in 2001. Current customers include pharmaceuticals giant Glaxo SmithKline.
But more important, perhaps, than its technology, is the experience that KXEN's co-founders can offer. KXEN was founded in 1998 by three highly regarded French entrepreneurs and researchers. Haddad, for example, was the founder of Metrologie, which, when he left it, had revenues of more than FF4.5 billion (€1bn).
Such experience will be invaluable if KXEN is to prosper in the competitive data modelling tools market.