Current multi-user extended reality (XR) systems enforce a symmetric reality where all users perceive an identical world, fundamentally limiting personalized interaction. A shared digital artifact cannot, for instance, smile back at one visitor while answering another’s question. We address this limitation by introducing Holo-Artisan, a novel system architecture designed to enable asymmetric personalized reality, where co-located users receive unique, real-time feedback from a shared digital entity. In our design, local edge computing nodes concurrently process real-time user data including pose, facial expression, and voice. Generative AI models then drive digital artworks (e.g., a volumetric Mona Lisa) to respond uniquely to each viewer. A cloud-assisted collaboration platform composes these interactions into a shared scene and renders high-fidelity, personalized views for glasses-free holographic displays. To preserve privacy and continuously improve personalization, we integrate federated learning (FL), allowing edge devices to locally fine-tune AI models and share only anonymized model updates. This edge-centric approach minimizes latency and bandwidth, ensuring a synchronized shared experience with individual customization. Through Holo-Artisan, static exhibits are transformed into dynamic, “living" artworks that engage each visitor in a personal dialogue, heralding a new paradigm for cultural heritage interaction.