Tech startup of the month.

• Author: Peterson, Eric
• Position: Hightech Coloradobiz - Valen Technologies Inc. - Company Profile

INITIAL LIGHTBULB: About 20 years before taking higher math to the insurance industry with Valen, Dax Craig and Richard Vlasimsky met as junior high school students in Tulsa, Okla. In 1993, Craig moved to Boulder to pursue an MBA. One of his school projects turned into a successful antenna company, Xertex Technologies, which he sold to market leader Centurion Wireless Technologies in 2000.

Meanwhile, Vlasimsky was working in software, as a consultant for Accenture and as a developer and product manager for other companies. After Craig fulfilled his commitment to Centurion in early 2002, he "wanted to get back into entrepreneurship," and gave old friend Vlasimsky a call. The pair had talked about starting a company together for a decade, and the time for a new venture--Valen--was right for both of them.

For CEO Craig, his involvement with Valen arose more from "opportunity than experience," while CTO Vlasimsky's know-how with artificial intelligence and data mining provided the basis for the company's technology.

IN A NUTSHELL: Valen's engineering team developed software to help companies make core business decisions, with the initial market being the insurance industry. The company's first product, Risk Manager, runs on an algorithm-based platform called Decision Studio (which will also underpin future Valen products for other industries).

"These algorithms are based on a branch of statistics known as statistical learning theory," explained Craig. Whereas traditional statistics presumes the answer to a question is known, "statistical learning theory actually assumes you don't know the answer. You fit an outcome to a model, rather than a model to an outcome."

An example as it pertains to Risk Manager: "An insurance company is guessing about risk when they underwrite a new client," said Craig. "(Underwriters) tend to focus on one or two variables. We actually take the entire set of potential risk characteristics--100, 200, 500 characteristics--and give the underwriter complete and total recall of...