Lenny's PodcastBrendan Foody: How Mercor hit $500M run rate in 16 months
Lawyers, bankers, and engineers design rubrics and RL environments for top AI labs; the work pays up to $500 an hour as an entirely new job category.
At a glance
WHAT IT’S REALLY ABOUT
Experts Writing AI Evals Are Powering History’s Fastest-Growing Companies
- Lenny Rachitsky interviews Brendan Foody, CEO and co-founder of Merkor, a labor marketplace that supplies expert talent to AI labs to write evals and post-training data, enabling rapid improvements in model capabilities.
- Merkor has gone from near-zero to roughly $400–500M in revenue run rate within 16–17 months, working with most top AI labs and major AI application companies, proving the scale of demand for high-quality human feedback and evaluation.
- Foody explains that we’re entering an “era of evals,” where defining and measuring what “good” looks like for models—across professional domains like law, medicine, and software—is the key bottleneck to progress, and a massive new category of work.
- They also discuss the future of labor markets, which skills will remain valuable, how founders can detect and pursue fast-moving market opportunities, and the cultural principles that enabled Merkor’s hypergrowth.
IDEAS WORTH REMEMBERING
5 ideasEvals are becoming as critical as the models themselves.
If the model is the product, evals are both the product requirements and the benchmark—defining what success looks like, enabling reinforcement learning, and serving as the sales collateral that proves model capabilities to the market.
The main bottleneck for top AI labs is expert human judgment, not raw data.
Labs increasingly need highly skilled professionals—lawyers, radiologists, bankers, engineers—to design rubrics, tests, and RL environments that evaluate and reward correct behavior in complex domains models can’t yet master.
A massive new job category is emerging around AI evaluation and post-training.
Contrary to the dominant narrative of job loss, AI is creating demand for tens of thousands (and growing) of experts who define tasks, write evals, and grade model outputs—often earning $95–$500 per hour on flexible, project-based work.
The most resilient careers will be in elastic-demand fields and AI power users.
Roles like software engineering, product management, and other high-leverage knowledge work will likely expand as productivity rises; individuals who learn to deeply integrate AI into their daily workflows will outperform their peers.
Founders should chase clear pull in fast-moving markets, not force fit.
Merkor’s hypergrowth came from noticing strong demand signals from top labs and incumbents’ weaknesses, then aggressively pivoting and focusing on that pocket instead of trying to push a slower, traditional hiring product.
WORDS WORTH SAVING
5 quotesIf the model is the product, then the eval is the product requirement document.
— Brendan Foody
The market is bound by the amount of things where humans can do something that models can't.
— Brendan Foody
Models are only as good as their evals.
— Brendan Foody (relaying a customer quote)
We grew from one to $400 million in revenue run rate in 16 months. Fastest ascent in history.
— Brendan Foody
You can just do stuff. So many people have ideas, but the barrier is initiative.
— Brendan Foody
High quality AI-generated summary created from speaker-labeled transcript.
Get more out of YouTube videos.
High quality summaries for YouTube videos. Accurate transcripts to search & find moments. Powered by ChatGPT & Claude AI.
Add to Chrome