Workforge creates scalable verified environments for task-driven ai agents
Scaling Verifiable Environments for Long-horizon Work Agents
Computation and LanguageArtificial Intelligence
Summary
Work agents are AI systems that perform complex professional tasks by interacting with digital files and tools, but creating environments for these agents is difficult and time-consuming. The authors introduce WorkForge, a method that builds realistic, large-scale digital workspaces using real-world files, automatically generating verifiable tasks based on the content. This approach makes it possible to train AI agents to handle long and complicated workflows more reliably. Their experiments show substantial improvements in AI performance across many job domains.
What this means in practice
- •For enterprise ai teams: Generate realistic, large-scale task environments from real business files to train AI systems for complex workplace roles.
- •For software development teams: Automate the creation of verifiable testing setups derived from actual project artifacts to improve reliability in AI-assisted software tools.
Authors
Jiazheng Zhang, Long Ma, Yunxian Yang, Zhiheng Xi, Zhikai Lei, Yajie Yang, Chenyang Liao, Enyu Zhou, Yang Nan, Yuchen Tian, Senjie Jin, Yibo Wang, Wei He, Boyang Liu, Jixuan Huang, Xin Guo, Zhezheng Hao, Xinbing Liang, Zhihao Zhang, Changzhi Zhou, Wiggin Zhou, Tao Gui, Qi Zhang, Xuanjing Huang, Clarenceai, Aiden Adams
Abstract
Work agents operate over digital artifacts to execute professional knowledge-intensive work, requiring training environments that support long-horizon interaction and trustworthy verification. However, hand-crafted environments incur prohibitive engineering overhead that prevents environment scaling, whereas synthesis methods sacrifice workspace complexity, realism, or grounded verifiability. To bridge this gap, we introduce WorkForge, a scalable synthesis framework for constructing verifiable work-agent environments from real-world resources. Starting from expert workflows, WorkForge first identifies the resources, decisions, and deliverables required by each workflow. It then retrieves relevant real-world files and organizes them into a workspace. WorkForge inspects the workspace to extract concrete, checkable facts about its content. These factual anchors fix which task types the workspace can support and how their outcomes can be verified. Therefore, WorkForge derives each task's instructions, solution plan, and complementary programmatic and semantic verifiers directly from these factual anchors, keeping verification traceable to observable workspace evidence. Furthermore, we construct 16.7K verifiable environments across 40 professional domains, with workspaces collectively covering 60 file types. Post-training Qwen3.5-35B-A3B-Base improves GDPVal from 45.5 to 73.6 and APEX Score from 5.0 to 21.3, while enabling Qwen3.5-27B to achieve highly competitive performance and outperform strong competitors. Our analyses confirm the efficacy of the proposed method and reveal consistent scaling behaviors across both data volume and interaction horizons.