For enhancing human-AI collaboration this paper presents the design and development of an emotion-aware and explainable chatbot. Simplified version of GoEmotions dataset were used to build a system, in which basic five primary emotional categories such as joy, anger, sadness, fear and neutral were consider from fine-grained labels. A simplified distillation of the DistilBERT model achieves efficient emotion classification with strong performance. For fostering transparency and user trust, a confidence scores for each detected emotion integrates an interpretability layer by chatbot in addition to prediction. A small AI agent was built that can understand emotions and explain its actions. This demonstrates a promising new technology for improving human–AI interactions in areas like healthcare, education, customer service, and other human-AI interaction domains. This work adds significant value to the broader field of explainable AI by demonstrating how efficiency, accuracy, and interpretability can be successfully combined. Also adding directions for future advancement through richer datasets and extended evaluation.