<p dir="ltr">The future of agentic AI lies in systems that engage users through socially intelligent, adaptive interactions. <i>Personified AI</i> refers to agents with distinct personalities, value systems, and communication styles, while <i>embodied AI</i> extends this with multimodal expression through voice, gesture, and behavior. These capabilities are essential for high-trust applications such as education, healthcare, and coaching.This paper presents IntelliExo, a modular AI advisor panel designed to simulate diverse, domain-specific personas for interactive guidance. As a work in progress, we also explore the potential integration of synthetic data to support advisor consistency and personalization, and propose future extensions toward embodied interaction via voice synthesis and avatar interfaces. These enhancements aim to increase realism, inclusivity, and engagement in personified AI systems.</p>