A multisensory environment (MSE) is a specialized room designed to present controlled stimuli to users, enhancing therapeutic experiences by stimulating multiple senses. Research suggests that introducing interactivity in these environments can amplify their benefits. However, MSEs are often expensive and located in specific facilities, limiting accessibility for many users. Virtual Reality (VR), with its highly interactive and immersive nature, presents a promising alternative for replicating MSEs in a more accessible and cost-effective way. This paper proposes a multi-user closed-loop architecture for developing a VR-based multisensory environment. The architecture is designed to process multimodal inputs, including electroencephalogram (EEG), electrocardiogram (ECG), and user behavioral data, enabling dynamic interactions within the VR application. These inputs trigger adaptive changes in the virtual environment, influencing user responses that are continuously monitored, thus creating a feedback loop. Additionally, the architecture integrates artificial intelligence for real-time processing and supports multi-user interactions, broadening the scope of VR applications in therapeutic settings, and, in this case, enhancing the potential of VR as an accessible alternative to traditional MSEs.