Mercor's AI Interviewer Has Vetted 300000 Job Candidates. Now It's ...

Mercor founders Adarsh Hiremath, Brendan Foody and Surya Midha are hoping to use AI to solve a "broken" hiring process for their peers.

Uly Inc. for Mercor Led by three 21-year-old Thiel Fellows, Mercor has raised $32 million from Benchmark, Peter Thiel and others for its AI-powered jobs marketplace. The startup’s revenue run rate is already in the tens of millions, growing 50% monthly — and profitable.

In August 2023, weeks after wrapping up their sophomore years of college at Georgetown and Harvard, Brendan Foody met with friends Adarsh Hiremath and Surya Midha to convince them to drop out. That January, the former highschool classmates had launched a side hustle called Mercor to match engineers in India with startups needing freelance coding help. In just a few months, they’d reached a revenue run rate of $1 million, raking in $80,000 in profit without any outside financial support.

Once Foody’s pitch was over, Midha turned to Hiremath, his partner in winning a national championship in policy debate two years before, to ask one more question: “Hey man, how hard could this really be?”

For Mercor’s founders, so far the answer seems to be “not very.” Their software, which uses artificial intelligence to vet and interview job candidates, then match them to open roles, has conducted more than 100,000 interviews, evaluating 300,000 people in less than two years. At still just 21 years old, Foody, Hiremath and Midha — who have all dropped out of school and become Thiel Fellows — now manage a business that’s still profitable on a run rate of tens of millions, growing 50% month over month.

And their 20-month-old startup is now valued at $250 million, following a $32 million Series A led by Benchmark and the firm’s newest partner Victor Lazarte, who has joined Mercor’s board. Billionaire investor Peter Thiel, Twitter cofounder Jack Dorsey and two OpenAI board directors, Quora CEO Adam D’Angelo and former Treasury Secretary Larry Summers, also personally invested.

“We were fascinated by labor, and how brilliant people all around the world aren’t getting opportunities,” explained Foody, Mercor’s CEO. “Applying large language models and automating the processes associated with talent assessment, we can solve this inefficiency.”

Initially a manual service to match and place temporary talent, Mercor’s marketplace now depends on its own large language model, building on top of the latest off-the-shelf options from shops like OpenAI and fine-tuned on its own proprietary data around its job-seeking process.

Applicants participate by uploading their resumes, then taking a 20 minute video interview with Mercor’s AI; about half that time is spent discussing the candidate’s experience, the other half responding to a relevant case study. With that, the seeker’s application is matched across all possible open jobs across Mercor’s marketplace; for more specialized roles, a second, tailored AI interview might follow.

For employers, Mercor promises to quickly connect qualified candidates who can flexibly scale up and down through contracted hourly, part-time and full-time commitments. Mercor’s largest pool of such talent remains India, although the U.S. is close behind. The roles can vary, too, including engineering and product development but also work in design, operations and content. Several leading AI labs, which Foody said he was unable to name, have used Mercor to seek out professional writers and videographers to create more data. Other employers have included companies in consulting, engineering, finance and law, the CEO said. Even Mercor itself relied exclusively on this talent until February of this year, though it now has 15 employees.

Mercor’s ability to predict applicant performance, trained against evaluations developed by hiring experts, already surpassed human recruiters in the startup’s tests, according to Foody, who said the startup will share its methodology in a published paper in November. Eventually, Mercor hopes to predict and automate longer-term performance reviews, too — helping employers track and cultivate talent over time. “There’s already immense economic value in predicting human ability better than a human… but adding the rest is where we believe the long-term defensibility and bigger economic value will occur,” Foody said.

Mercor isn’t Foody’s and Hiremath’s first business. Years before he dropped out of Georgetown, Foody managed a small business building websites for startups while attending Bellarmine College Preparatory in San Jose. Harvard-bound Hiremath, whom Foody knew through competing on the school’s elite debate team, helped with the coding; Midha, Hiremath’s friend from years before, also became close with Foody, becoming his roommate at Georgetown.

For their policy debate competitions, the founders spent months researching positions before arguing both sides. Similar discussions about the job market — particularly unfair in the developing world, Midha and Hiremath, both the sons of Indian immigrants, felt — led them to spin up Mercor with the help of an engineering contractor they hired remotely in India. “It felt so deeply wrong that hiring and recruiting processes arbitrarily tossed out half of applicants based on resume, or not having prestigious work experience, or certain types of education history,” said Midha. “Building a more meritocratic way to hire is the largest lever on people's overall quality of life.”

After one month spent in New York, Mercor’s founders moved to San Francisco to work on the startup full-time last fall. As CEO, Foody focused on product and sales; Midha, as COO, its operations and enterprise contracts; Hiremath, as CTO, its engineering and machine learning functions. In January, they announced a $3.6 million seed round led by General Catalyst; in March, all three were named Thiel Fellows, granting them $100,000 and access to a network of other founders and tech executives as friends and potential customers.

Mercor’s profit margins were widening, so it had no real interest in funding when Benchmark’s Lazarte met Foody later in 2024. His experience bootstrapping games startup Wildlife Studios led Foody to take one meeting; the offer of a helicopter ride piloted by Midas List investor Peter Fenton scored Benchmark another. Former Midas List fixture Bill Gurley, the semi-retired former Uber investor, met Mercor next as Benchmark fended off other firms to preempt Mercor’s funding round. (Not before its founders, leaning into their debate experience, walked away over the VC’s proposed valuation at least once.)

“In venture, we take a lot of meetings, but every now and again you take one that is like: ‘Wow, this is really impressive,’” said Lazarte. “I knew this guy [Foody] was really special. And the company’s vision to help match people with their best job, I think it’s a noble one.”

That sentiment will prove a tougher sell if businesses utilize Mercor to downsize headcount and rely more on outsourced contractors, or to help AI research labs train models that put other specialists out of work. Benchmark is counting on Mercor to help create better jobs for people, not speed up AI taking them all away, Lazarte argued. Mercor will also solve the more complex challenge of matching workers to full-time, salaried jobs over time, he added.

Mercor faces some well-capitalized competition in the talent market such as startup unicorn Andela, which recently hired an Uber executive as its new CEO; traditional outsourcing shops might also turn to off-the-shelf AI tools to look to compete. Another challenge: maintaining fresh interest on both sides of Mercor’s marketplace. The startup has conducted targeted outreach on LinkedIn for certain job postings, Foody conceded, but he insisted that most new users come in on their own, utilizing resources like resume feedback and interview practice available on its site.

Success for Mercor, Foody said, means “building a unified global labor market,” one that can learn the ideal job for every person, then provide it to them. Midha put it more simply: “Getting a billion people jobs, that would make me really happy.”