Mercor

Mercor is a platform that connects highly specialized human experts with frontier artificial intelligence labs and enterprises to generate high-quality training data, conduct reinforcement learning from human feedback (RLHF), and train AI agents at scale. The company focuses on organizing human intelligence—specifically the deep knowledge of professionals across medicine, law, finance, science, and management—to improve the reasoning, accuracy, and safety of advanced AI models. The platform functions as a two-sided marketplace: on one side, leading AI labs and enterprises post tasks that require expert judgment, such as reviewing model outputs, creating gold-standard datasets for supervised fine-tuning, evaluating agent behavior, or generating complex reasoning chains. On the other side, carefully vetted experts—including internal medicine specialists, hematologists/oncologists, biology PhDs, litigation lawyers, private equity professionals, management consultants, and venture capitalists—apply for these roles and are compensated at rates ranging from $80 to $180 per hour. The system processes daily payouts for completed tasks and has already created thousands of roles, with hundreds of experts hired recently in each category. Mercor’s product suite includes APEX (a general platform for human intelligence tasks), APEX-Agents (for training and evaluating AI agents that perform autonomous actions), APEX-SWE (focused on software engineering tasks), and APEX Research (supporting cutting-edge scientific research). These tools allow AI companies to outsource the most challenging validation and data generation work to domain specialists, rather than relying on generalist annotators. For example, a medical AI model being developed for diagnosis might need hundreds of board-cert