Аннотация:This study presents an agent-based multimodal information extraction system designed for nanomaterials research. The proposed framework integrates natural language processing, computer vision, and domain ontologies to extract structured knowledge from scientific publications, patents, and databases. By combining text and image analysis, the system enables automatic recognition of material compositions, synthesis methods, and properties. The results demonstrate the potential of agent-based architectures for accelerating knowledge discovery in nanoscience and supporting data-driven materials design.