Inspiration
Navigating digital relationships can be confusing and emotionally draining — especially when you're unsure how the other person feels. From overthinking to anxiety around responding "the right way," we realized there's no tool designed to support emotional clarity in texting. People often get stuck in toxic relationships and can't understand the warning signs.
What it does
HeartByte allows users to upload screenshots of text conversations and receive: A real-time analysis of how engaged or emotionally invested the other person seems. Sample responses based on the tone and topic of the conversation, aimed at reducing anxiety and improving communication. A chatbot trained to offer thoughtful advice and empathetic guidance, promoting self-worth and mental clarity.
How we built it
Frontend: Built using VS Code and hosted via Lambda Cloud. Backend: Python-based architecture to handle OCR preprocessing, sentiment analysis, and classification. We imported libraries like PIL and Pytesseract AI Models: We imported OpenAI and wrote calls to it. We finetuned it slightly to better answer the prompts for our purposes.
Challenges we ran into
We faced huge challenges with the HTML as neither of us has ever coded in that before. It took a long time and a lot of trial and error to sync up our backend code with our frontend. Another challenge we faced was extracting the text and categorizing it from screenshots. We found that the color fluctuated a lot of the messages so we had to adjust the color thresholds to classify the texts as incoming or outgoing correctly.
Accomplishments that we're proud of
We gained a lot of HTML knowledge and debugging experience after this project. We are proud that our frontend looks so amazing!
What we learned
It is our first time working in depth with AI, so we learned a lot about importing and finetuning it. We also learned frontend development because the most we had done with that is by using no-code platforms.
What's next for HeartByte
We hope to further train our AI model to provide more in depth analysis. We also hope to transform it into a mobile extension so people can access it right from their Messages app. We can make a dashboard to track compiled relationship health over time to further support people's mental health in their relationships and advise when it is time to leave.
Log in or sign up for Devpost to join the conversation.