SerenityAI

Inspiration

We were inspired to create SerenityAI to address the growing need for accessible mental health support, helping users overcome barriers to seeking help.

What It Does

SerenityAI is an AI-powered mental health companion that offers personalized support, including emotional check-ins, coping strategies, and tailored resources for managing stress and anxiety. It has a built-in chatbot trained to answer your personal queries and act as your personal therapist. SerenityAI’s chatbot processes the input using an LLM trained on therapeutic responses.

How We Built It

We developed SerenityAI using natural language processing and machine learning, collaborating with mental health experts to ensure safe and effective content while focusing on user-friendly design. For the front-end we used React, Node and Tailwind CSS, which we integrated with our Python backend through the Flask framework and APIs. We also leveraged Databricks technologies such as LanceDB, as a vector store for dynamic context retrieval, and Delta Lakes, for structured conversation history maintenance and token limit optimization.

Challenges We Ran Into

We faced challenges in creating empathetic AI responses, balancing complexity with simplicity, and addressing privacy concerns to protect user data.

Accomplishments That We're Proud Of

We were able to deliver on our vision to provide users with accessible mental health care. We worked through various challenges in development but were ultimately successful in implementing a user-friendly platform, working on top of an optimized RAG pipeline, while using Databricks technologies in a creative way.

What We Learned

We learned the ins and outs of the software development life cycle in a collaborative environment. We were also able to expand our knowledge in various technologies and tools and how to utilize them in an optimal way. Ultimately, our team was able to grow through this valuable experience.

What's Next for SerenityAI

We hope to make the platform more human-like by implementing audible, face-to-face conversations, providing a more comfortable setting for users to share their thoughts. We plan to optimize these conversations by training models on facial expressions and sentiment analysis, to understand the users emotions and provide more accurate and relevant responses. We could also explore partnerships with mental health organizations, and enhance our AI algorithms for better personalization. Additional LLM fine-tuning and advanced analytics implementations for personalized therapy sessions based on user behavior leave options to improve SerenityAI looking bright.