Zephyr

Zephyr is an interactive journal for mental health support that provides affirmations and recommendations tailored to the user's journal entries.

Comment

Inspiration

As students who are constantly balancing classes, assignments, and extracurriculars, we often find ourselves stressed and overwhelmed. Research shows that journaling can help reduce anxiety and promote healing [https://www.webmd.com/mental-health/mental-health-benefits-of-journaling], and it has helped a lot of us personally grapple with our stressors. One thing that we noticed was that while journaling helped us recognize our emotions, it was still hard to find the next steps of what to do when we were feeling how we did. This is why we chose to create Zephyr which turns our journal entries into positive affirmations and recommended podcasts and songs. This way, we can take the next steps best for our mental health without having to put in extra time and research ourselves.

What it does

Zephyr takes in a user's journal entry, and based on the content of their entry, will provide them with five affirmations relevant to their situation, a song recommendation to help them process the emotion they are feeling, an activity recommendation to help them channel their energy towards something productive, and a podcast recommendation that can help them with more specific advice on how to grapple with their current situation. Everything is tailored to the user's journal input, not hardcoded.

How we built it

We built the bot that processes the journal entries in nlx and uses integrated Anthropic API. We built our front-end website completely using streamlit and backend with Google Cloud and mongodb for local storage.

Challenges we ran into

It was initially very difficult to figure out how to get nlx, which is a platform new to all of us, to process the entries in the way we wanted it to. We spent a lot of time troubleshooting and changing the flow of our bot so that we got the outputs we wanted using AI instead of hardcoding them.

Accomplishments that we're proud of

We are proud of our final bot and the website we've developed. It took us a long time to integrate all the intents we wanted into one bot, and we're happy with how effective it is in providing relevant affirmations and recommendations. We also learned something new which was building a website completely using Python which was cool.

What we learned

We learned a lot about nlx and how APIs work. None of us have really worked with AI at all before and had never used programs like nlx, so we definitely gained a lot of knowledge on how chatbots and low-code platforms work from this experience.

What's next for Zephyr

We hope to scale the responses Zephyr can provide based on journal entries. We also want to implement a streak tracking feature so users are encouraged to update their journals daily. A voice conversation feature is also something we wanted to integrate but were not able to due to the time constraints of this hackathon; that way, users could converse with the journal by speaking instead of through text. Also, we are planning to add a memories tab that connects with Google Cloud and fetches and displays previous entries for the user to go through.

Built With

Share this project:

Updates