Charles Law found the prospect of looking for a job daunting. After more than 38 years in the U.S. Air Force Reserve and then on active duty, Law, now a master sergeant, was retiring and looking for work in the civilian world. But it had been 19 years since he’d last searched for a job outside the military.
“It was kind of scary,” Law says of his employment search. “First trying to decide what it is that I wanted to do next and then having to deal with the civilian sector after having been attached to the military for so long.”
Law says he found the jargon used by corporate employers confusing. “The terminology and lingo is just not the same,” he says. “And trying to get an understanding of what they are looking for and trying to help them understand what skills I had and how that relates to a civilian job, that is a bit challenging.”
Each year, about 200,000 members of the armed forces leave the military and transition to civilian life. And many of them face the same struggles Law did. But starting next year, a new software system based on artificial intelligence will help many of these veterans find work.
Eightfold AI, a four-year-old company based in Mountain View, Calif., that makes A.I.-enabled recruiting and “talent management” software, won a U.S. Department of Labor competition to build a software system to help improve its existing Transition Assistance Program (TAP), which helps those leaving the military prepare for civilian life, including finding employment. The startup beat out 50 other companies, including more established players in recruitment technology such as LinkedIn, Vantage Point Consulting, and even JobPath Partners, a company that specializes in helping companies hire veterans.
To win the competition, Eightfold created a website for military members that takes a U.S. military job code and translates that job into its component skills—but also any related skills the job seeker might have, says Dan Hopkins, senior director of Eightfold’s public sector business. “If I am in artillery, what is the equivalent in civilian life?” he says. “But that person will have a lot of training and leadership skills.”
Eightfold is among dozens of startups—including Pymetrics, Mya Systems, and HireVue—that are applying A.I. to recruitment. The company’s software can predict what skills the job seeker might be able to learn based on what that person already knows. “If you know skill A, might you also know skill B or could you learn skill C, and how quickly could you learn it,” Hopkins says.
The system Eightfold built can also assess what rank and grade a person holds and how quickly they were promoted to that rank—which can be a good proxy for the ability to learn new skills and leadership abilities. “If a candidate is an E-6 and they made that in five years—that is actually very good and means they advanced quickly. All that information is lost in a conventional résumé; it isn’t conveyed,” he says.
Kamal Ahluwalia, Eightfold’s president, says veterans have many soft skills that aren’t picked up on a normal résumé, such as the ability to work in ambiguous environments. “They have this in spades, and they can work under pressure,” he says.
The problem with the government’s existing TAP program is that it is largely based on a self-assessment questionnaire about skills that the job seeker completes, combined with keyword searches for open positions, Hopkins says. But many veterans sell themselves short in terms of the skills they actually have, he says, and keyword searches can return an overwhelming number of potential jobs, with no good way to screen them to find the best opportunities. For instance, a combat engineer who did a search of the TAP’s job database for open engineering roles just in the Washington, D.C., area in October would have to weed through more than 22,000 possible roles.
By contrast, Eightfold’s software curates a list of specific job openings for a candidate. It also curates job candidates for employers. And, for servicemen and -women leaving the military, it curates a network of veterans in similar roles or closely related ones that job seekers can tap.
“For transitioning service members it is really important to have a relevant network—with an emphasis on quality and context,” Hopkins says. “In this case we are surfacing other veterans in the role that I’m a strong match for.”
Eightfold’s victory in the competition came with $720,000 in prize money and was also supported by the Defense and Veterans Affairs departments.
Law, the retiring master sergeant who helped Eightfold test its software, says having that connection with another veteran at prospective employers is often a decisive factor. “I think when I find that person in the company, there is a sense of comfort,” he says. “We as a military, we feel we are all a family whether we have known that person one day or several years. I do find that they have a better understanding of where you are coming from.”
Eightfold’s A.I. has been trained on more than a billion employee profiles that the company harvested from various public and proprietary data sets, which include public announcements of new hires in trade journals and posts from sources such as alumni magazines, along with information on how those people were promoted over time, Ahluwalia says. It used this data to teach its A.I. algorithm about more than 500,000 different job titles and what the 1.4 million underlying skills inherent in those jobs are. The algorithm then uses this information to try to forecast the likely career progression of any given job seeker.
The company has been backed by more than $180 million in venture capital funding since its founding, including a $125 million financing round announced last month that was led by venture capital firm General Catalyst, based in Cambridge, Mass., which valued the startup at $1 billion.
This story has been corrected. A previous version incorrectly stated Dan Hopkins, senior director of Eightfold’s public sector business, was a veteran.