disable drop cap

While waiting to meet Sheldon Gilbert for lunch in front of a Midtown Manhattan restaurant recently, I'm a little surprised to see him walk right by me. Not that we've ever met before. It's just that the supremely savvy and laid-back 32-year-old can predict, with alarming accuracy, the future. Or, rather, he can help you, if you have an e-commerce Web site—like Barneys, one of his clients—predict when a customer will buy something and for how much.

Sci-fi fantasy? No, more like mighty fine science. Sheldon, who grew up on the island of St. Lucia and goes by Shel to those close to him, is founder and CEO of the tech start-up Proclivity Systems (formerly Gilbert Systems). "Human behavior is predictive," he says with a faint Caribbean accent and bright smile. "If it weren't, we wouldn't be in business." The company's flagship offering and big HAL soothsayer is called Proclivity, a predictive engine—remember the olden days when search engines were all the rage?—that launched earlier this year. Everyone who travels along the Internet leaves a digital trail, and Proclivity is a method that makes sense of the madness. For businesses, it means marketing on steroids and dramatically increased sales. But it's doubtful that anyone, even Proclivity, could have anticipated the eclectic route that led Gilbert to become one of the young pioneers ushering in the next generation of the Web.

For starters, he wanted to be a physician. He graduated from Yale with a degree in molecular biochemistry and biophysics without ever having taken a computer course. (Anne Wojcicki, Sergey Brin's wife and cofounder of the genetic profiling company 23andMe, was there at the same time.) He then spent two years conducting genetic research at Cornell University Medical Center and Rockefeller University. "Since I was a kid, I've been interested in understanding the systems behind these seemingly random things in nature," Gilbert says. But this was the late nineties, when the tech bubble was being inflated with tornado-gauge force, and a couple of his buddies from Yale were starting a dot-com, which he joined as a consultant. The company, like so many others, went bust, but it was an important lesson for Gilbert in how e-commerce transactions worked. More crucially, he also picked up computer programming—by teaching it to himself after work each night. (Some people have happy hour; others find comfort in zeroes and ones. These days Gilbert can't even find the time or, evidently, the right speed, for a girlfriend: "In two weeks they're usually going 50 mph—marriage, kids—and I'm like, 'Hmm, I'm trying to figure out if we should see 3:10 to Yuma at 7:00 or 10:00.' ")

Spurred on by the intriguing similarities between the problem-solving methods he used in the genetic lab and those he applied to programming puzzles, Gilbert took a job as director of business intelligence at the online fashion retailer bluefly.com. It was there that the idea for Proclivity began to come to him in a series of "Aha!" moments. Realizing he had something big on his hands, he left his job, deferred enrollment in New York University's graduate program in computational genomics, and, in the bedroom of his Harlem loft, spent a year writing the algorithm ("architecting information" is how he describes it) that is now the patent-pending basis for Proclivity. "Graduate school will always be there," he reasons.