Imagine effortlessly building and deploying machine learning models, even without coding expertise. That’s the promise of AutoML, but traditional AutoML systems can be complex and difficult to use. Enter AutoML-Agent, a groundbreaking new framework that leverages the power of large language models (LLMs) to automate the entire machine learning pipeline, from data retrieval to model deployment. AutoML-Agent acts like your personal AI assistant, understanding your needs through natural language instructions. It then coordinates a team of specialized LLM agents, each an expert in a different stage of the AutoML process. One agent handles data preprocessing, another focuses on model selection and tuning, and another manages the deployment—all working in concert like a well-oiled machine. This multi-agent approach allows AutoML-Agent to explore a wider range of solutions, optimizing for both accuracy and efficiency. But unlike other LLM-based AutoML systems, AutoML-Agent doesn't rely on fixed templates or predefined workflows. Instead, it employs a “retrieval-augmented planning” strategy, drawing on the vast knowledge embedded within LLMs and external sources like research papers. This allows AutoML-Agent to adapt to various tasks and constraints, ensuring it always uses the most up-to-date techniques. AutoML-Agent also features a multi-stage verification process to catch errors and refine the generated code, ensuring you get a high-quality, deployable model. In experiments, AutoML-Agent outperformed existing AutoML methods across diverse tasks, demonstrating its adaptability and effectiveness. While AutoML-Agent marks a significant step toward democratizing AI, challenges remain. Future improvements could focus on reducing its reliance on specific LLM backbones and extending its capabilities to even more complex tasks like reinforcement learning. But the vision is clear: AutoML-Agent paves the way for a future where anyone can harness the power of AI, regardless of their technical background.