Medical Information functions are exploring artificial intelligence (AI) and advanced analytics to improve the speed, consistency, scalability, and intelligence of Medical Inquiry management. Generative AI can support intake, information extraction, classification, scientific-content retrieval, evidence synthesis, response drafting, documentation, and analytics. These opportunities arise within a controlled environment in which scientific accuracy, response scope, source traceability, privacy, safety reporting, product-quality escalation, documentation, and human accountability are essential. This paper proposes Compliance by Architecture: a practitioner-informed model in which controls are embedded across the complete inquiry pathway rather than concentrated at a final review gate. The framework was developed through a practitioner-informed conceptual synthesis of professional Medical Information guidance, peer-reviewed literature, pharmacovigilance and electronic-record requirements, and responsible-AI principles and frameworks. The model comprises seven interconnected layers: channels and intake; identity, context, and reference data; governed scientific content and knowledge; AI orchestration and retrieval; workflow, routing, and the system of record; analytics and monitoring; and human decision and accountability. It also introduces an automation boundary that distinguishes AI-accelerated activities, human-reviewed activities, and human-owned decisions or escalations. The framework integrates established Medical Information practice, pharmacovigilance expectations, electronic-record controls where applicable, and responsible-AI principles while remaining explicitly conceptual rather than regulator-issued. Its intended contribution is a practical architecture for deploying AI as a governed scientific-workflow capability, not as an autonomous scientific authority. Future empirical work should test the framework across jurisdictions, inquiry types, content models, and human-AI workflows.