ARCHIVES

This paper introduces AGENTIC AI, an autonomous multi-agent framework developed to simplify the machine learning workflow with minimal human involvement. The system employs specialized agents for data analysis, feature engineering, model training, evaluation, explainability, and report generation within a collaborative architecture. It supports multiple learning algorithms and incorporates hyperparameter optimization to improve predictive accuracy. SHAP-based methods are used to provide model interpretability, while a semantic memory layer enables knowledge reuse across executions. Implemented with a FastAPI backend and a React-based interface, the framework allows users to upload datasets, visualize results, and access generated reports. Experimental results show that AGENTIC AI offers an efficient, scalable, and interpretable solution for automated machine learning development.