Supporting graduate EFL proposal writing through a staged NotebookLM-supported workflow

Graduate L2 students frequently experience difficulty initiating and sustaining research proposal writing, often manifesting as observable drafting disruption. This Research into Practice paper reports on a staged instructional workflow implemented in a graduate EAP writing course in Taiwan, where selected NotebookLM features were integrated as pedagogical scaffolds to support proposal development. The workflow was organized around four writing stages: planning, monitoring, reflecting, and revising, and incorporated explicit guardrails for responsible AI use. Data documenting the practice included proposal drafts, weekly learning logs, reflective essays, and an end-of-course survey. Drawing on classroom evidence, the paper illustrates how instructor-designed drafting prompts supported task initiation, Audio Overviews (auto-generated podcasts) facilitated comprehension and re-engagement, and study guides, FAQs, and briefing documents aided synthesis and organization. The practice illustrates how AI-supported scaffolding helped students sustain writing momentum and develop clearer proposal structures. Implications are discussed for EAP teachers seeking to integrate generative AI tools into graduate writing instruction in pedagogically purposeful ways.