Toward Multimodal, Memory-Augmented Agents: Just-in-Time Interfaces for eXtended Reality

This doctoral research investigates multimodal, memory-augmented agents for Extended Reality (XR) to address current deficiencies in integrating generative AI, such as Large Language Models (LLMs), for improving the interaction through multimodality, personalization, and adaptivity. The research wants to investigate how to synthesize just-in-time, context-aware interfaces and integrate persistent memory systems. The primary objective is to identify models and technological architectures enabling the development of proactive, embodied XR assistants capable of mitigating cognitive load and enhancing user interaction through dynamic user interface synthesis, memory architectures, and anticipatory agent behaviors.