product · May 21, 2026
Mercor AI Training Firm 2026 Study: Every Top-Tier AI Agent Failed Most of 480 Workplace Tasks
Share the canonical public link.
Mercor researchers evaluated agents from top-tier models on 480 workplace tasks in a 2026 study. Every agent failed to complete most of its duties according to the findings. The study benchmark tested specific functions like software engineering and other workplace scenarios. Mercor is referenced as the AI training firm behind the evaluation published in the January 22 2026 TechCrunch report. The Fortune article dated May 18 2026 cites the Mercor study in context of AI agent adoption limits at companies like ClickUp.
Below validation threshold — auto-passed without scoring