Aethorix v1.0: An Integrated Scientific AI Agent for Scalable Inorganic Materials Innovation and Industrial Implementation

Yingjie Shi, Yiru Gong, Yiqun Su, Suya Xiong, Jiale Han, Runtian Miao

arXiv:2506.16609·cs.CE·Published 2025-06-19·Updated 2025-11-17

Artificial Intelligence (AI) is redefining the frontiers of scientific domains, ranging from drug discovery to meteorological modeling, yet its integration within industrial manufacturing remains nascent and fraught with operational challenges. To bridge this gap, we introduce Aethorix v1.0, an AI agent framework designed to overcome key industrial bottlenecks, demonstrating state-of-the-art performance in materials design innovation and process parameter optimization. Our tool is built upon three pillars: a scientific corpus reasoning engine that streamlines knowledge retrieval and validation, a diffusion-based generative model for zero-shot inverse design, and specialized interatomic potentials that enable faster screening with ab initio fidelity. We demonstrate Aethorix's utility through a real-world cement production case study, confirming its capacity for integration into industrial workflows and its role in revolutionizing the design-make-test-analyze loop while ensuring rigorous manufacturing standards are met.

TopicsGenerative Design & Molecule Optimization, Quantum Chemistry & Force Fields

Tagsab-initio drug-discovery generative-model mlip

arXiv categoriescs.CE

arXiv abstract pagePDF