Increasingly, companies are looking to predictive recommendation software to match employees to mentors. These programs—which use the same type of predictive technology as Spotify, Amazon, and Netflix—match an employee to a mentor on the same type of data points that match potential partners on dating sites like eHarmony. “People didn’t necessarily have access to someone who could be their ideal mentor,” says Mike Bergelson the co-founder and CEO of one such program, called Everwise. “We wanted to reinvent mentoring.” Large employers, including Walmart, Genentech, and Oracle have begun using recommendation software, with some of these companies paying upwards of $125 per employee per month to manage their online mentorship relationships. “It’s the future of HR to draw on this kind of data,” says John Boudreau, a management professor at the University of Southern California. However, relying on data to predict the success of a human relationship is not without its critics. Harry Reis, a psychology professor at the University of Rochester last year published a paper challenging these predictive algorithms. “The biggest factor in whether a relationship succeeds or not might be called chemistry—how well two people connect with each other," he wrote, "and no one has yet come close to providing credible scientific evidence that this can be quantified with an algorithm."