No Priors Ep. 110 | With Mercor CEO and Co-Founder Brendan Foody
Sarah Guo (host), Brendan Foody (guest), Elad Gil (host)
In this episode of No Priors, featuring Sarah Guo and Brendan Foody, No Priors Ep. 110 | With Mercor CEO and Co-Founder Brendan Foody explores aI Evals, Labor Displacement, and Building a Global Talent Marketplace Mercor CEO Brendan Foody explains how the company uses large language models to evaluate and match talent, initially for AI data-labeling and now broadly for high-skill roles. He argues that models already outperform most human hiring managers on text-based assessments and that evals for economically valuable work—especially agentic, tool-using tasks—are becoming a core bottleneck. The conversation explores power-law performance in knowledge work, how reinforcement fine-tuning (RFT) enables data-efficient customization of agents, and why building evals may become the most common knowledge job. They also discuss rapid job displacement, the future mix of humans and AI agents in a unified global labor market, and what skills remain worth cultivating as AI advances.
AI Evals, Labor Displacement, and Building a Global Talent Marketplace
Mercor CEO Brendan Foody explains how the company uses large language models to evaluate and match talent, initially for AI data-labeling and now broadly for high-skill roles. He argues that models already outperform most human hiring managers on text-based assessments and that evals for economically valuable work—especially agentic, tool-using tasks—are becoming a core bottleneck. The conversation explores power-law performance in knowledge work, how reinforcement fine-tuning (RFT) enables data-efficient customization of agents, and why building evals may become the most common knowledge job. They also discuss rapid job displacement, the future mix of humans and AI agents in a unified global labor market, and what skills remain worth cultivating as AI advances.
Key Takeaways
AI models already outperform most human hiring managers on text-based assessments.
Mercor finds that LLMs can more accurately predict job performance from resumes, interviews, and written work than typical human managers, especially in high-volume, comparable roles, making it increasingly irrational not to rely on model recommendations.
Get the full analysis with uListen AI
The human data market has shifted from low-skill crowd work to elite eval creation.
Early AI training relied on cheap, low-skill annotation; now the frontier is recruiting highly capable domain experts who can design nuanced evals and work closely with researchers to push model capabilities via RL and RFT.
Get the full analysis with uListen AI
Evals for economically valuable, agentic work are the main bottleneck to automating knowledge jobs.
Most benchmarks test narrow, zero-shot questions, but real jobs involve coordination, tool use, and end-to-end outcomes; building rich, job-like evals—especially for agents—is essential to turn existing model capability into automation.
Get the full analysis with uListen AI
Verifiable domains (math, code, clearly gradable tasks) will be automated fastest.
Where outputs can be checked automatically or against clear criteria, models can rapidly improve through reinforcement learning; harder domains have sparse, noisy signals (e. ...
Get the full analysis with uListen AI
Labor displacement in many white-collar roles is likely to be fast and politically painful.
Foody anticipates significant near-term disruption in functions like customer support and recruiting, with broader sectors following; he expects populist backlash and a need to rethink wealth redistribution and human roles.
Get the full analysis with uListen AI
Building evals may become one of the most common knowledge work jobs.
As every economically valuable task needs evals and feedback loops, Foody predicts huge demand for people—often existing employees—to codify “what good looks like” so agents can learn each domain, organization, and workflow.
Get the full analysis with uListen AI
The long-term vision is a global unified labor market of humans plus AI agents.
By collapsing manual matching costs with software, Mercor aims to let companies hire from a global pool and candidates access global opportunities; over time, Foody expects humans and agents to compete and collaborate in the same marketplace.
Get the full analysis with uListen AI
Notable Quotes
“We train models that predict how well someone will perform on a job better than a human can.”
— Brendan Foody
“For everything that we want LLMs to be good at, we need evals for those things.”
— Brendan Foody
“I think displacement in a lot of roles is going to happen very quickly and it's going to be very painful.”
— Brendan Foody
“It would not surprise me if [creating evals] becomes the most common knowledge work job in the world.”
— Brendan Foody
“It makes way for a global unified labor market that every candidate applies to and every company hires from.”
— Brendan Foody
Questions Answered in This Episode
If AI models outperform human managers at hiring, how should laws and regulations adapt to govern automated evaluations and prevent abuse or bias?
Mercor CEO Brendan Foody explains how the company uses large language models to evaluate and match talent, initially for AI data-labeling and now broadly for high-skill roles. ...
Get the full analysis with uListen AI
What concrete steps can workers in vulnerable roles (e.g., customer support, recruiting) take now to transition into higher-resilience, AI-leveraged careers?
Get the full analysis with uListen AI
How can organizations design evals that genuinely capture complex traits like taste, judgment, and leadership rather than just easily gradable proxies?
Get the full analysis with uListen AI
In a global unified labor market mixing humans and AI agents, what mechanisms will ensure fair competition, wages, and access to opportunity across countries?
Get the full analysis with uListen AI
At what point does continued human input into evals and RL training actually make superhuman models worse, and how will we detect that tipping point?
Get the full analysis with uListen AI
Transcript Preview
(instrumental music) . Hi, listeners, and welcome to No Priors. Today, we're chatting with Brendan Foudy, co-founder and CEO of Merkle, the company that recruits people to train AI models. Merkle was founded in 2023 by three college dropouts and Thiel fellows. Since then, they've raised $100 million, surpassed 100 million in revenue run rate, and are working with the top AI labs. Today, we're talking about where the data for foundation model training will come from next, evaluations for state-of-the-art models, and the future of labor markets. Brendan, welcome to No Priors. Brendan, thanks so much for doing this.
Yeah, thanks for having me. Excited to be here.
So you guys have had a, like a wild last six months or so.
Mm-hmm.
Um, there's huge traction in the company. Can you just talk a little bit about, uh, what Merkle does?
Yeah, so at a high level, we train models that predict how well someone will perform on a job better than a human can. So similar to how a human would review a resume, conduct an interview, and decide who to hire, we automate all those processes with LLMs. And it's so effective, it's used by all of the top AI labs to hire thousands of people that train the next generation of models.
What are the skills and like job descriptions that the labs are looking for right now?
It, it's really everything that's economically valuable because reinforcement learning is becoming so effective that once you create evals, the models can learn them and how to, uh, you know, improve capabilities. And so for everything that we want LLMs to be good at, we need evals for those things. Um, and it ranges from consulting to software engineers, all the way to hobbyists and video games and, and everything that you can imagine under the sun. And it's really whatever capabilities you're seeing the foundation model cap- companies invest in, or even application layer companies invest in, uh, the evals are upstream of all that.
Uh, and are you also helping companies outside of the core foundation models with this similar type of hiring? Or is it mainly just focused on AI models right now?
Yeah, so actually when we started the business, it was totally unrelated to human data. It was just that we saw that there were phenomenally talented people all around the world that weren't getting opportunities and we could apply LLMs to make that process of finding them jobs more efficient. And then we realized after, uh, you know, meeting a couple of customers in the market that there was just this huge vacuum because of the transition in the human data market. And that the human data market used to be this crowdsourcing problem of how do you get a bunch of low- and medium-skilled people that are writing barely grammatically correct sentences for the early versions of ChatGPT. And it was transitioning towards this vetting problem of how do you find some of the most capable people in the world that can work directly with researchers to push the frontier of model capabilities. But we've still kept that core DNA of hiring people for roles, human data and otherwise. Um, and a lot of our customers hire for both.
Install uListen to search the full transcript and get AI-powered insights
Get Full TranscriptGet more from every podcast
AI summaries, searchable transcripts, and fact-checking. Free forever.
Add to Chrome