This paper presents an AI-driven interactive assistant system that combines face detection, natural language processing (NLP), and animated 3D characters to enhance human-computer interaction. Utilizing a fine-tuned Llama language model for generating context-aware responses, the system dynamically adapts to user expressions detected in real-time using CNN-based and Haar cascade-based facial recognition. A Unity-based 3D avatar visually represents responses through lip-syncing and emotional expressions. This integration bridges the gap between traditional text-based chatbots and emotionally aware virtual assistants. Applications span customer support, elderly care, and education, with potential future enhancements including emotion-sensitive responses, gesture recognition, and voice cloning.