This paper introduces AtomAgents, a physics-aware multi-agent artificial intelligence system designed to address complex materials design challenges through collaborative AI agents. The system integrates large language models (LLMs), multi-modal data processing, physics-based simulations, and comprehensive result analysis to enable autonomous alloy design and discovery. AtomAgents leverages the strengths of multiple AI agents working together in a dynamic environment to solve complex materials design tasks, demonstrating its ability to autonomously design metallic alloys with enhanced properties compared to their pure counterparts. The system's core components include a planning module, a perception module that integrates multi-modal data, and an action module that executes decisions based on the planning module's guidance. AtomAgents can process diverse data formats, including text, images, and tabular data, and integrate knowledge from various sources such as academic literature, databases, and simulations. The system is capable of performing atomistic simulations, including molecular dynamics and density functional theory, and can generate and validate new hypotheses through these simulations. The paper presents several experiments demonstrating the effectiveness of AtomAgents in solving complex tasks in alloy design and analysis. These experiments include material property calculations, analysis of screw dislocation core structures, and multi-scale mechanics problems involving fracture toughness. The results show that AtomAgents can accurately predict key characteristics of alloys and highlight the role of solid solution alloying in developing advanced metallic alloys. The system also reduces the need for human intervention, enabling autonomous design workflows and making advanced simulations more accessible to non-experts. The multi-agent system's ability to integrate diverse data modalities and perform complex reasoning tasks makes it particularly suited for addressing the challenges of materials design. The system's collaborative approach allows for iterative refinement of strategies, merging insights across disciplines to progressively evolve towards optimal solutions. AtomAgents demonstrates the potential to significantly enhance the efficiency and effectiveness of complex multi-objective design tasks, opening new avenues in fields such as biomedical materials engineering, renewable energy, and environmental sustainability. The integration of physics-aware AI with multi-agent collaboration represents a significant advancement in materials science, enabling more efficient and effective material design and discovery processes.This paper introduces AtomAgents, a physics-aware multi-agent artificial intelligence system designed to address complex materials design challenges through collaborative AI agents. The system integrates large language models (LLMs), multi-modal data processing, physics-based simulations, and comprehensive result analysis to enable autonomous alloy design and discovery. AtomAgents leverages the strengths of multiple AI agents working together in a dynamic environment to solve complex materials design tasks, demonstrating its ability to autonomously design metallic alloys with enhanced properties compared to their pure counterparts. The system's core components include a planning module, a perception module that integrates multi-modal data, and an action module that executes decisions based on the planning module's guidance. AtomAgents can process diverse data formats, including text, images, and tabular data, and integrate knowledge from various sources such as academic literature, databases, and simulations. The system is capable of performing atomistic simulations, including molecular dynamics and density functional theory, and can generate and validate new hypotheses through these simulations. The paper presents several experiments demonstrating the effectiveness of AtomAgents in solving complex tasks in alloy design and analysis. These experiments include material property calculations, analysis of screw dislocation core structures, and multi-scale mechanics problems involving fracture toughness. The results show that AtomAgents can accurately predict key characteristics of alloys and highlight the role of solid solution alloying in developing advanced metallic alloys. The system also reduces the need for human intervention, enabling autonomous design workflows and making advanced simulations more accessible to non-experts. The multi-agent system's ability to integrate diverse data modalities and perform complex reasoning tasks makes it particularly suited for addressing the challenges of materials design. The system's collaborative approach allows for iterative refinement of strategies, merging insights across disciplines to progressively evolve towards optimal solutions. AtomAgents demonstrates the potential to significantly enhance the efficiency and effectiveness of complex multi-objective design tasks, opening new avenues in fields such as biomedical materials engineering, renewable energy, and environmental sustainability. The integration of physics-aware AI with multi-agent collaboration represents a significant advancement in materials science, enabling more efficient and effective material design and discovery processes.