Recently, Large language models (LLMs) have been used to generate various content. However, a fundamental limitation is that LLMs only work with a single data modality (i.e., text); Large Multimodal Models (LMMs), on the other hand, fill that gap by allowing other input modalities (e.g., image, video), thus adding context to complete the task successfully. This research aims to investigate the usability of LMMs for network management and operations using an agentic workflow approach. This paper shows the overall concept and preliminary experiments.