Abstract
Artificial intelligence (AI) provides new opportunities to improve the efficiency, accessibility, and multilingual reach of Islamic da'wah. This study developed and evaluated AI-Dakwah, an AI-supported workflow for producing Friday sermon texts in Indonesian, English, Javanese, and Arabic. The study employed a Research and Development (R&D) design with qualitative and experimental elements through an Adaptive AI Development Framework covering preparation, topic exploration, text generation and translation, Islamic-content validation, expert review, editing, and digital publication. The development process involved a needs analysis with ten preachers and fifty congregants and a dataset of 40 Friday sermon texts in four languages. Large Language Models were used to prepare initial drafts, while generated materials were checked against Qur'an and hadith resources and reviewed by two Islamic scholars and one Arabic-language expert. User evaluation showed positive results, with mean ratings of 4.4/5 for language quality, 4.0/5 for material depth, and 4.7/5 for conformity with Islamic values. Reported preparation time decreased from approximately eight hours manually to two hours with AI assistance. The findings indicate that AI-Dakwah can support more efficient and accessible multilingual sermon production when AI is used as an assistive technology under human and scholarly supervision. The study also identifies limitations related to regional-language data, evaluation design, and the absence of controlled behavioral comparisons. The study concludes that responsible human-in-the-loop governance is essential for maintaining theological accuracy, cultural relevance, and editorial accountability in AI-assisted Islamic da'wah.