International Journal of Scientific Research and Engineering Development

Modern tutoring platforms tend to struggle with three recurring problems: they sometimes give wrong information, they aren't tied to what's actually being taught in a specific classroom, and they pile extra work onto already-busy teachers. SAHAYAK was built to address all three of these at once. It's an educational platform that pulls content directly from textbooks, gets teachers to review and approve that content, and then uses it to tutor students in a way that actually matches their curriculum. The whole thing runs on a multi-agent setup, with four agents handling different jobs — gathering data, storing information, helping students, and keeping everything coordinated. Google Gemini and OpenAI APIs power the language side of things, while teachers and students interact through a clean Next.js interface that also handles approvals and tracks student progress. Under the hood, the system reads textbook PDFs, breaks them into meaningful chunks, turns them into teacher-approved knowledge clusters using RAG, and builds out personalized study paths and quizzes from there. When tested, it hit a faithfulness score of 0.98 — which basically means it almost never makes things up, a big step up from general-purpose AI tools. The project shows that AI can be useful in classrooms without sacrificing accuracy or teacher control.

Prof. Ashwini M C, Maheshwari M, Hisham ul hakeem A, Maanya J, Prajwal P G,"SAHAYAK: A Teacher-Aligned Multi-Agent Tutoring System for Curriculum-Grounded Student Support Using Retrieval-Augmented Generation" International Journal of Scientific Research and Engineering Development, V9(3): Page(823-828) May-June 2026. ISSN: 2581-7175. www.ijsred.com. Published by Scientific and Academic Research Publishing.