Samarth: An AI-Powered Multi-Agent System for Automated and Compliant Individualized Education Program Generation

Authors:

Dr. Sowbhagya M P, Sudhanva Joshi, Sravya Illuri, Vikas K B

Abstract—Samarth is an AI-powered multi-agent sys-tem designed to automate the generation of Individual-ized Education Programs (IEPs) for students with special needs, addressing a significant challenge in India’s inclu-sive education framework under the RPwD Act, 2016. By leveraging a dynamic “Thought–Action–Observation” loop within a LangGraph-based multi-agent architecture, the platform integrates advanced large language models and Retrieval-Augmented Generation (RAG) to aggre-gate, analyze, and synthesize diverse student data from clinical, academic, and behavioral sources. Samarth’s agents collaboratively ensure regulatory compliance and produce personalized, structured, and high-quality IEPs through iterative reasoning. The human-in-the-loop de-sign maintains educator oversight while automating up to 85% of documentation workload, enabling educators to prioritize direct student engagement. This scalable and ethical solution supports digital health and accessible ed-ucation goals, especially in under-resourced regions, im-proving educational outcomes for students with disabili-ties.

Keywords—Individualized Education Program (IEP), Multi-Agent System, Artificial Intelligence (AI), Lang-Graph Framework, Retrieval-Augmented Generation (RAG), Special Education, Rights of Persons with Disabili-ties (RPwD) Act, Educational Technology, Human-in-the-Loop, Inclusive Education.