This study aimed to describe students’ lived experiences in utilising the Gemini application as Artificial Intelligence (AI) in Arabic language learning and to identify the essence of these experiences. This study employed a qualitative phenomenological approach involving five students selected purposively based on their experience using Gemini for Arabic language learning. Data were collected through in-depth interviews, observations, and documentation and analysed through data reduction, data display, and conclusion drawing. The findings show that students experienced Gemini as a supportive learning medium in translating Arabic texts, searching for vocabulary (mufradat), understanding nahwu and sharaf, constructing Arabic sentences, and completing academic assignments. Students experienced greater ease and perceived efficiency in learning due to quick access to explanations provided by Gemini. However, they also experienced limitations when some responses were inaccurate or did not correspond with the learning context. These experiences led students to verify Gemini’s responses through lecturers, textbooks, dictionaries, classical Islamic texts and other academic sources. The essence of students’ experiences was identified by examining recurring meanings emerging from the interview, observation, and documentation data, particularly the experiences of convenience and perceived learning efficiency, along with the need for critical verification of AI-generated content. Thus, Gemini is experienced as a supplementary learning medium rather than a primary source of knowledge, supporting students’ learning while maintaining critical thinking, information verification, and academic responsibility skills.