Automation in community pharmacy is often discussed as though medication tasks can be transferred to digital systems whenever adequate technical performance is demonstrated. This framing overlooks differences in clinical consequence, uncertainty, contextual dependence, patient vulnerability, error detectability, and the possibility of correcting an inappropriate action. This article develops an original, non-empirical delegation-boundary model for determining when automation may proceed, when professional review is required, and when delegation should stop. Delegability is defined as a conditional property of a specific medication-task instance rather than an inherent characteristic of an entire service, technology, or occupational role. The proposed model assesses six interacting dimensions: clinical consequence, ambiguity, reversibility, contextual dependence, patient vulnerability, and detectability of error. These dimensions inform three conditional states. Proceed permits bounded automated execution under defined safeguards; Escalate requires review by an authorized professional before action; and Stop retains human control because necessary context, recovery capacity, accountability, or safety conditions are insufficient. The architecture also includes task specification, uncertainty communication, professional authority, monitoring, incident review, and boundary reassessment. Application is intended to distinguish automatable components from judgment-dependent components within dispensing, verification, counselling, and follow-up rather than classify whole services as automatable or non-automatable. Empirical evaluation would require task-specific technical validation, workflow simulation, human-factors assessment, medication-safety analysis, patient and equity evaluation, external validation, and lifecycle monitoring. The model is a proposed conceptual synthesis, not a clinical recommendation, legal standard, validated scoring system, or deployment-ready framework. Its principal contribution is to reposition pharmacy automation as a conditional delegation decision requiring evidence about both system performance and the consequences of transferring professional work.