The increasing complexity and intricacy of deep neural networks (DNNs) pose significant challenges in understanding and interpreting their decision-making processes. As these models grow in size and sophistication, there is a growing need for methods that facilitate the comprehension and analysis of their internal mechanisms. This necessity has led to the rise of specialized techniques collectively referred to as Explainable AI (XAI), which aims to improve model transparency and interpretability. However, most existing XAI methods rely on static visualizations, providing fixed insights into model behavior without capturing the dynamic nature of learning processes, such as gradient updates, weight adjustments, and feature importance evolution over time. This study presents LogosXAI, an emerging framework designed to extend XAI capabilities by integrating both static and dynamic visualization techniques. LogosXAI enables real-time tracking of neural activations, signal propagation, and gradient flow, allowing for a deeper understanding of how deep learning models process information. Additionally, the system explores a hybrid approach, combining symbolic rule-based explanations with deep learning interpretability to illustrate the complementary roles of white-box and black-box methods. The study further discusses the potential of virtual reality (VR) as an innovative tool for immersive AI explainability, enabling users to navigate DNN architectures in 3D and interactively analyze decision-making processes.