Inspiration
As programmers, many of us have a rubber duck hanging around on our desks, especially when we're trying to painfully debug our code. This cute companion sits silently as we explain our thoughts out loud to better understand the problems we're trying to solve. Inspired by this simple yet effective practice, we wanted to expand it to be a conversational mental-health tool, offering support and interaction beyond just listening. In addition, the recent rise of chatbots has demonstrated how AI can assist with communication and analysis. We saw this as an opportunity to leverage generative AI to help users engage in meaningful, customized conversations and guide them through processing their emotions. Lastly, Teddy bears are symbols of comfort and friendliess, which is why we decided to theme our project around that.
What it does
HeyTed is a tabletop companion designed to be a supportive presence for users who need to vent or express their thoughts. With a warm and empathetic personality, Ted listens patiently, offering reassurance and understanding in times of distress, without trying to "solve" the user's issues.
While Ted can be used alone, this projet can be especially helpful to those who are currently seeking mental health support. At the end of each conversation, Ted provides a thoughtful summary, highlighting key points from the discussion. In addition, Ted proposes next steps for the user to encourage healthy ways to maintain their mental health. This way, users are able to gather their thoughts and share critical points to their healthcare providers.
HeyTed can also be used as a standalone mobile app, and supports text conversations as well.
How we built it
For our app, we used Django for the front and back end, and the Gemini API to send and receive responses. We also used Google Speech Recognition for text-to-speech functionality. The UI was designed on Figma and for the physical component, we used servo motors to animate the stuffed animal's arms.
Challenges we ran into
We ran into a number of technical bugs. For example, we had issues with setting up our environment, getting errors related to cookies, inability to generate a response from Gemini, and bypassing web security checks in our external devices. We had issues with integrating the hardware with the rest of the code.
Accomplishments that we're proud of
We are proud of the UI design, mastering a new web framework (Django), and being able to integrate various components software and hardware components together.
Especially for many of us, Python isn't a language we frequently use for applications, so challenging ourselves to work with it and using different libraries was a difficult but rewarding experience.
This was also our first time integrating AI into a project, so learning how to effectively engineer prompts and end up with such a cute chatbot was a bit accomplishment.
What we learned
We were all inexperienced with Python, so this proved to be quite challenging. We had to learn how to prompt engineer, how to integrate APIs with the rest of our code, and also how to work together as a team.
What's next for HeyTed
- Turning the responses of the user into usable information for health care providers and mental health practitioners.
- Securely storing multiple chat sessions and analyzing trends of user's emotions.
- Redesigning the hardware so it connects directly to the mobile app.
Log in or sign up for Devpost to join the conversation.