How to build a communications AI assistant

Abstract

This workshop presentation outlines a practical strategy for small alumni relations offices to build a custom, no-code AI communications assistant using tools like NotebookLM. Facing the reality of limited personnel (one manager and rotating interns) who must adapt daily university news across five digital platforms, small offices frequently experience a "consistency gap" and fragmented institutional tone. To bridge this gap, the author proposes feeding an AI assistant with a "walled garden" of institutional memory, including style guides, FAQs, and top-performing past publications. This implementation delivers an 80% efficiency shift by automating heavy structural drafting and multi-platform cross-posting (LinkedIn, X, Instagram, Facebook, and blogs), allowing the core team to focus on strategy, empathy, and the final human touch. The framework breaks down the process into four structured blocks: defining the anatomy of the institutional voice, flipping institutional news into alumni-centric stories, mapping core knowledge sources, and designing the precise prompt architecture for passive onboarding of new team members.