EnterpriseBench Corecraft: Training Generalizable Agents on High-Fidelity RL Environments

TL;DR

Training AI agents in high-fidelity enterprise environments like CoreCraft enhances their ability to generalize across diverse, real-world tasks, with models showing significant improvements and transfer to out-of-distribution benchmarks.

Contribution

Introduction of CoreCraft, a realistic enterprise RL environment, and demonstration that training in such environments improves generalization of AI agents beyond training tasks.

Findings

Models trained in CoreCraft improve task pass rates from 25.37% to 36.76%.

Transfer gains of +4.5%, +7.4%, and +6.8% on out-of-distribution benchmarks.

Environment properties like diversity and realism are key to enabling generalization.

Abstract

We show that training AI agents on high-fidelity reinforcement learning environments produces capabilities that generalize beyond the training distribution. We introduce CoreCraft, the first environment in EnterpriseBench, Surge AI's suite of agentic RL environments. CoreCraft is a fully operational enterprise simulation of a customer support organization, comprising over 2,500 entities across 14 entity types with 23 unique tools, designed to measure whether AI agents can perform the multi-step, domain-specific work that real jobs demand. Frontier models such as GPT-5.2 and Claude Opus 4.6 solve fewer than 30% of tasks when all expert-authored rubric criteria must be satisfied. Using this environment, we train GLM 4.6 with Group Relative Policy Optimization (GRPO) and adaptive clipping. After a single epoch of training, the model improves from 25.37% to 36.76% task pass rate on held-out evaluation tasks. More importantly, these gains transfer to out-of-distribution benchmarks: +4.5% on BFCL Parallel, +7.4% on Tau2-Bench Retail, and +6.8% on Tool Decathlon (Pass@1). We believe three environment properties are consistent with the observed transfer: task-centric world building that optimizes for diverse, challenging tasks; expert-authored rubrics enabling reliable reward computation; and enterprise workflows that reflect realistic professional patterns. Our results suggest that environment quality, diversity, and realism are key factors enabling generalizable agent capabilities.