fb-pixel Skip to main content

Cambridge startup wants to fill the ‘skills gap’ with AI assistance

AdeptID’s software aims to help those without college degrees find jobs.

AdeptID cofounders Fernando Rodriguez-Villa and Brian DeAngelis raised $3.5 million in seed funding to develop software for matching people's skills with job openings.AdeptID

Cambridge startup AdeptID, formed last year to build an AI-backed talent-matching service for companies, announced a seed funding round on Thursday.

The eight-person company raised $3.5 million in a deal led by Zeal Capital Partners and including Better Ventures, Boston nonprofit JFF’s Employment Technology Fund, and other angel investors.

It’s a small amount of money for a big idea. AdeptID’s software is designed to help companies find workers who have the proper skills for a job opening even if they lack the seemingly required experience or credentials, such as a college degree.

Cofounders Fernando Rodriguez-Villa and Brian DeAngelis met at agriculture tech startup Indigo Ag, which acquired Rodriguez-Villa’s previous startup, TellusLabs, in 2018. The pair both “had enough of a screw loose required to go and start something new,” Rodriguez-Villa, now CEO of AdeptID, said. DeAngelis, who has a doctorate in computational neuroscience, is the company’s chief data scientist.

“Tens of millions of folks in the US labor market have a really hard time transitioning between jobs, particularly between industries, which results in a number of folks that are not just unemployed, but under-employed or employed in sectors in structural decline,” Rodriguez-Villa said. “We wanted to see if there was a technology application that could start to address that social problem.”


Many people have learned important skills, such as how to deal with customers or pay attention to small details, in retail jobs, for example, that could make them qualified for roles in higher-paying fields that are having trouble filling openings, such as pharmacy technicians.

The software also can help companies in emerging industries like renewable energy, where few job candidates would have previous experience as a wind turbine technician, for example. “You can’t look for people who’ve done that job before, because that job hasn’t existed long enough,” Rodriguez-Villa said.


The technology that matches people’s skills with suitable jobs is not unlike the software that Netflix uses to recommend movies or Stitch Fix uses to recommend outfits, he explained.

The system can also recommend additional training that might be needed for candidates lacking the full set of skills required for a job.

The artificial intelligence was trained by looking at actual career paths and applicant pools of thousands of workers (with personal information removed) to uncover skills that might be transferable from one job to another.

It’s a good time to be raising money for an HR-related startup. The pandemic led to waves of layoffs and then a struggle to fill open positions in many industries, prompting employers to look for new, digital ways to recruit, hire, and perform other HR functions. Startups in HR technology raised $10.5 billion so far in 2021, more than was raised in the prior three years combined, according to PitchBook. In the Boston area, other HR tech startups include workplace analytics developer Humanyze, employee engagement startup WeSpire, and talent platform Catalant.

Not all will succeed. Biased algorithms have become a critical failing of some AI software in the field — Amazon canceled development of an AI-based recruiting app a few years ago after discovering it favored men over women.

AdeptID said its system has been designed to avoid the pitfalls that plague other AI-driven systems. And instead of using people’s individual characteristics to eliminate them from consideration, AdeptID’s software takes the opposite approach and tries to find reasons why someone might be qualified for a role.


“Any company that is using ‘AI’ or data science to train models based on outcomes has to be careful about the susceptibility of those models to bias,” Rodriguez-Villa said. “That’s something we’ve prioritized early on and have included in the structure of the algorithms themselves.

“Just because someone’s a cashier doesn’t mean we’re going to throw them out,” he added. “We’re actually going to look for the parts of being a cashier that are relevant.”

Aaron Pressman can be reached at aaron.pressman@globe.com. Follow him on Twitter @ampressman.