We introduce a novel machine learning framework called Agentic Automated Machine Learning (AutoML), designed to integrate advanced agentic capabilities with traditional AutoML workflows. Agentic AutoML extends the machine learning model development lifecycle by incorporating Large Language Models (LLMs), with statistical analysis, model training, and hyperparameter tuning, etc., all tailored to a simple user experience. The Agentic AutoML builds on standard AutoML machine learning algorithms, including tree-based methods such as random forests and gradient boosting, as well as formula-driven models such as polynomial regression and artificial neural networks, also equipped with automatic statistical analysis, sensitivity assessments, error analysis, and automated report generation. Agentic AutoML architecture offers real-time guidance throughout the entire project workflow, enhancing user interaction and decision-making. By simplifying the application of machine learning algorithms, Agentic AutoML significantly broadens accessibility for users with varying levels of programming skills and machine learning expertise, making it a valuable tool for diverse user groups.