Relaxo

Welcome Relaxo - an app for your mental health! It includes a chatbot that's there to comfort you in any situation. Relaxo also analyzes your daily journal entries to recognize any patterns over time.

Comment

Inspiration

Many people worldwide suffer from mental health issues every day. Whether it's anxiety, stress, depression, or even burnout, everyone can use a friend. However, not everyone is surrounded by good communities. People also like to be comforted in many different ways. Hence, I built Relaxo.

What it does

Relaxo is an all-in-one app that strives to improve mental health over time. It contains a chatbot, Relaxo, who is always there to comfort you.

Additionally, the app has a journal. Upon posting a journal entry, generative AI analyzes it's sentiment and classifies if your day went positively, negatively, or neutrally.

Finally, the app has a page dedicated to breathing exercises. You simply describe how you're feeling in a few words, and AI is used to choose an optimal breathing exercise for you. Animations tell you when to inhale/exhale so that you can easily follow along.

How I built it

  • I used OpenAI's gpt-3.5-turbo Chat Completion model for all of the AI-related features. This includes the chatbot, sentiment analysis, and a breathing technique chooser. I connected this to a small API built with Python & FastAPI.
  • I used React on the front end to develop a user interface.
  • I used Firebase to handle user authentication
  • I used Netflix's DGS Spring Boot & GraphQL framework to build an API with a PostgreSQL database on the back end to handle storing user information and journal entries.

Challenges I ran into

  • The chatbot was acting in unexpected ways at times - for example, when asked to metaphorically do something, the chatbot would respond with something like "I am an AI language model and am not able to..." To overcome this, I researched prompt engineering on DeepLearning.AI, which taught me how to create effective system messages for the model. I also added examples for the bot to learn from and consequently deliver better results.

Accomplishments that I'm proud of

  • I'm happy that I was able to connect Generative AI to solve a real-life problem that affects those around me. Because this was my first time working with OpenAI, I'm glad that I was able to quickly learn this technology and integrate it with technology I'm already familiar with (such as React, Spring, etc.).

What I learned

  • This was my first project working with OpenAI's models. I learned how to use the OpenAI library and its chat model. I also learned a little bit about prompt engineering and how it can help the results of an LLM.

What's next for Relaxo

  • I want to continue to utilize AI so that Relaxo can be able to analyze a full journal history and tell you patterns from the data - For example, do certain names in your journal entries correlate with negative days? Which months do you feel better/worse? Patterns/trends like these can significantly help identify problems and improve one's mental health.
  • I also want Relaxo to have a larger variety of breathing techniques that it can help you with. I'd love to create more advanced & unique animations for each exercise, too.
Share this project:

Updates