Researchers introduced WorkForge, a scalable synthesis framework for constructing verifiable work-agent environments from real-world resources. This enables trustworthy verification of long-horizon work agents, such as professional knowledge-intensive work, with a significant improvement in performance on a large-scale AI model.