The paper introduces AtomAgents, a physics-aware multi-agent AI platform designed to accelerate the design and discovery of new alloys. The platform integrates the capabilities of large language models (LLMs) and multi-modal data integration, enabling autonomous collaboration among AI agents to solve complex materials design tasks. AtomAgents addresses the limitations of existing data-driven models by leveraging the dynamic collaboration among agents with expertise in various domains, including knowledge retrieval, multi-modal data integration, physics-based simulations, and comprehensive results analysis. The platform enhances the efficiency of complex multi-objective design tasks and opens new avenues in fields such as biomedical materials engineering, renewable energy, and environmental sustainability. The paper presents several experiments demonstrating the effectiveness of AtomAgents in solving complex tasks, including material property calculation, screw dislocation core structure analysis, multi-scale mechanics problems, and hypothesis generation and validation. The results highlight the platform's ability to integrate diverse data modalities, automate and optimize the entire workflow, and reduce the need for manual intervention. The authors discuss the limitations and future perspectives of their approach, emphasizing the potential for further advancements in materials science through the integration of advanced deep learning models and generative tools.The paper introduces AtomAgents, a physics-aware multi-agent AI platform designed to accelerate the design and discovery of new alloys. The platform integrates the capabilities of large language models (LLMs) and multi-modal data integration, enabling autonomous collaboration among AI agents to solve complex materials design tasks. AtomAgents addresses the limitations of existing data-driven models by leveraging the dynamic collaboration among agents with expertise in various domains, including knowledge retrieval, multi-modal data integration, physics-based simulations, and comprehensive results analysis. The platform enhances the efficiency of complex multi-objective design tasks and opens new avenues in fields such as biomedical materials engineering, renewable energy, and environmental sustainability. The paper presents several experiments demonstrating the effectiveness of AtomAgents in solving complex tasks, including material property calculation, screw dislocation core structure analysis, multi-scale mechanics problems, and hypothesis generation and validation. The results highlight the platform's ability to integrate diverse data modalities, automate and optimize the entire workflow, and reduce the need for manual intervention. The authors discuss the limitations and future perspectives of their approach, emphasizing the potential for further advancements in materials science through the integration of advanced deep learning models and generative tools.