An AI-enhanced Production-Oriented Approach (POA) framework is proposed to improve English as a Foreign Language (EFL) writing instruction in higher education. By aligning POA’s motivating–enabling–assessing cycle with large language models such as GPT-4, the framework facilitates task-based, feedback-driven, and adaptive learning experiences. Implemented through a design-based research methodology in a university writing course, the framework led to measurable improvements in students’ writing performance, engagement with formative feedback, and learner autonomy. The results suggest that AI functions as an instructional augmentation rather than a replacement, offering scalable support for individualized scaffolding and functional feedback. The work contributes to smart language pedagogy by bridging pedagogical theory with AI capabilities and connecting human-led instruction with intelligent learning support.