Enterprise AI faces a reality check as agentic systems struggle with complex planning

Recent research from ServiceNow introduces EnterpriseOps-Gym, a high-fidelity benchmark revealing that current AI agents struggle with long-horizon planning, persistent state changes, and strict access controls in realistic enterprise environments.