Abstract-Future building automation will require complex (human-like) perception and decision-making processes not being feasible with classical approaches. In this paper, we address both the perception and the decision-making process and present an alerting model that reacts to perceived situations in a building with decisions about possible alerts. Perception is based on the neuro-symbolic information processing model, which detects candidate alerts. Integrated with perception, decision-making is based on the rule model of RuleML, which computes alerts to relevant building occupants about current opportunities and risks. A general model of neuro-symbolic alerting rules is developed and exemplified with a use case of building alerts.