The use of artificial intelligence (AI) technology in education continues to develop rapidly, yet not all regions can equally benefit from it, especially the 3T areas (frontier, outermost, and disadvantaged) in Indonesia. This research is motivated by concerns over this disparity and aims to understand how AI-based learning systems can be contextually adopted in schools facing infrastructure and resource limitations. Using a mixed methods approach, the study combines quantitative data from questionnaires distributed to 150 respondents (mostly teachers) and qualitative data from in-depth interviews with key informants who have direct field experience. Findings show that despite the willingness to adopt new technologies, obstacles such as unstable internet connectivity, lack of teacher training, and limited institutional support remain major challenges. Based on these findings, a flexible implementation model of an AI-Driven Tutoring System was developed, featuring a semi-offline format and supported by practical training materials tailored to local needs. This study not only contributes to the theoretical understanding of technology adoption in education but also offers practical implications for formulating more inclusive and adaptive policies. By integrating technical and social insights, this research promotes the development of educational technology that is more equitable, relevant, and responsive to local contexts.