This document records a methodology for integrating large language models into academic research and writing workflows, developed through extensive use on linguistics projects including corpus analysis, theoretical development, and book-length academic writing. The account covers: the philosophical stance toward LLMs as collaborative intellects rather than oracles; practical strategies for prompt design, iterative refinement, and multi-model deployment; validation protocols for empirical claims and citations; division of labour between human expertise and model capabilities; and specific techniques for drafting, structural reorganization, and stylistic refinement. The document is intended as a pedagogical resource for researchers and instructors developing AI literacy curricula. It reflects practice as of late 2025, using models including Claude (Anthropic), ChatGPT (OpenAI), Gemini (Google), Grok (xAI), and Kimi (Moonshot).