Due to the increase in depression cases and people’s reluctance to seek medical therapy, it has been an alarming situation to develop a chatbot that offers therapy in an accessible and easy manner so that individuals suffering from depression do not feel uncomfortable during social interactions with medical professionals. This research’s main objective is to present a web-based intelligent chatbot that offers therapy to users depending on the emotions shown in their conversations. The emotion detection part utilizes models such as BERT and RoBERTa coupled with CNNs to correctly categorize emotion in the user’s message, and the chatbot is integrated with Generative AI for giving personalized responses to the user. For testing the model we chose the ISEAR dataset that included 7666 sentences categorized into 7 major emotions that are joy, guilt, anger, disgust, fear, sadness, and shame.