✨ About the Project: Comfort-Buddy Comfort-Buddy was inspired by the idea that everyone deserves a small moment of joy, especially on tough days. Sometimes, all we need is a comforting word, a warm meal, a cozy vibe, or the right song to feel a little better. We wanted to create an emotionally intelligent assistant that could offer all four — instantly.
💡 Inspiration
This project was born out of our team's curiosity: "What if your mood could automatically generate the perfect comfort combo: support, food, colors, and music?" We imagined an app that blends mental wellness, emotional support, and generative AI — not to diagnose or treat, but simply to soothe.
🛠️ How We Built It
Frontend: Streamlit for rapid, interactive UI. LLM Backend: Groq API with llama3-8b-8192 model for ultra-fast mood-based responses. Image API: Unsplash API to generate recipe images based on dish names. Deployment: Hosted on Hugging Face Spaces. Languages/Tools: Python, Markdown, HTML/CSS (via Streamlit), Pandas for journaling. TextBlob — For sentiment analysis (detecting user mood from text)
🧠 What We Learned
Prompt engineering for emotionally intelligent responses. Integrating multiple APIs in a cohesive app (Groq + Unsplash). Session management and caching in Streamlit. Designing for user empathy and not just functionality.
😓 Challenges We Faced
Styling limitations of Streamlit — had to use custom CSS hacks. Ensuring image relevance and fallback when Unsplash queries failed. Managing prompt formats so that LLM output remains structured. Debugging environment issues on Hugging Face Spaces (e.g., missing CSS files).
🧪 Bonus: A Bit of LaTeX
To reflect our emotional model, we imagined: moodComfort=f(Mood,Empathy)+Recipe warm +Vibe color+emoji +Song mood A mix of psychology, art, and technology.
Log in or sign up for Devpost to join the conversation.