Camily: A BERT-Powered Chatbot for Employee Workplace Wellness

Mental health at the workplace is an emerging issue, yet the traditional methods of psychological approaches in the workplace fail mainly because they have limited accessibility, carry a stigma, time-bound, and lack direct applicability to the workplace environment. Chatbots provide a ready alternative offering 24/7 access, anonymity, scalability, and AI-driven personalization toward clients. It relieves stress immediately, conducts mental health screening or assessment, and intervenes using stress management reminders. Beyond access and practicality, chatbots are cheaper and stigma-free compared to traditional methods. Camily is a unique chatbot application that employs BERT (Bidirectional Encoder Representations from Transformers) for sentiment analysis. By using the BERT methodology, one was able to achieve as high as 96 percent accuracy through improved contextual word understanding and emotional tone detection in recognizing anxiety or stress levels from a conversation. Designed with a unique dataset of 1380 entries marked out with clear wellness lexicons, Camily provides immediate wellness tips, stress management techniques, or referrals when necessary. After chatting, it will predict the stress level to be Low, Medium, or High based on the chat and classify it from the dataset used. This research will complement a developing mental health support system that can be integrated into workplaces.