Designing DARTS: A unified AI system for tutoring, classroom interaction, assessment, and students? success

Building on the conceptual foundation presented in the first article on the evolution of Mastery Learning to DARTS (Sarkar, 2025), this second paper presents the design and technical architecture of DARTS (Dynamic Academic Response and Tutoring System).DARTS is an AI-powered, mobile-first Intelligent Tutoring System (ITS) designed to operationalize Mastery Learning at scale by integrating real-time classroom data with personalized, SMS-based tutoring.This study addresses key research questions concerning the practical and pedagogical potential of mobile-enabled ITS platforms, particularly in enhancing accessibility, responsiveness, and equity in learning environments.The paper details DARTS' core system architecture, including its feedback loop, AI-powered decision engine, natural language processing (NLP) layer, and its innovative use of behavioral classroom signals (such as attendance, quiz participation, and engagement) to trigger timely, individualized instruction.Each classroom workflow for attendance, quizzing, and brainstorming, is described alongside the system's diagnostic and intervention processes.The design draws upon Bloom's 2 Sigma model and advances in AI to transition DARTS from concept to a deployable system.While this paper focuses on the design and implementation of the system, it does not assess instructional outcomes.That analysis will follow in the third paper of this three-part series, which will present empirical classroom findings.The DARTS architecture builds on insights from the author's doctoral research and teaching experience and is currently protected by two pending U.S. patents (Application Nos.18/521,928 and 18/433,800).