Do Enterprise Systems Need Learned World Models? The Importance of Context to Infer Dynamics

World models enable agents to anticipate the effects of their actions by internalizing environment dynamics. In enterprise systems, however, these dynamics are often not fixed but defined by tenant-specific business logic that changes over time, causing learned world models to fail under deployment shift. We ask a question the world-models literature has not addressed: when the rules can be read at inference time, does an agent still need to learn them? We argue, and demonstrate empirically, that in settings where transition dynamics are configurable, runtime discovery is a more robust foundation for agent prediction than offline training. We propose enterprise discovery agents, which recover relevant transition dynamics at runtime by reading the system's configuration rather than relying solely on internalized representations. We introduce CascadeBench, a reasoning-focused benchmark for enterprise cascade prediction that adopts the evaluation methodology of World of Workflows on diverse synthetic environments, and use it together with deployment-shift evaluation to show that learned world models perform well in-distribution but degrade sharply as dynamics change, whereas discovery-based agents remain robust by grounding their predictions in the current instance. Our findings suggest that, in enterprise environments, agents should be designed to discover dynamics at runtime rather than rely on fixed world models