Inspiration As a team deeply passionate about accessibility and mental health, we recognized how difficult it can be for individuals with disabilities—especially children—to clearly express their emotions. Whether due to autism, Down syndrome, or other conditions, emotional expression is often misunderstood or overlooked. We wanted to build something that gives a voice to feelings when words are hard, and helps caretakers, especially parents, better understand and support their loved ones. Thus, MindMirror was born—a simple, empathetic, and accessible tool that translates emotions into clear, kind summaries.
What it does MindMirror allows users to select their emotional state and intensity using a friendly emoji slider. Once submitted, the input is sent to our backend where Gemini (Google's powerful generative AI) interprets and generates a simple, supportive summary. This summary is crafted with empathetic, accessible language tailored for children or individuals with disabilities, and displayed back to the user or parent. It’s emotion made visible and understandable—bridging the gap between feeling and support.
How we built it Frontend: Built using TypeScript, React, and Tailwind CSS, we designed a clean and accessible UI with an emoji-based slider to represent different moods and intensities.
Backend: We used Python and FastAPI to handle incoming data, process it using the Gemini API, and return the AI-generated summaries.
AI Integration: The backend securely loads a Gemini API key from environment variables, builds a custom prompt based on the user’s selected emotion and intensity, and parses the Gemini response for display.
Communication: The frontend sends user input to the backend using POST requests via fetch calls, receiving clean and clear summaries in return.
Challenges we ran into Integrating Gemini with the frontend: Ensuring that mood and intensity inputs could be correctly parsed, formatted, and interpreted by Gemini required careful prompt engineering and error handling.
API setup and security: Managing environment variables and ensuring the Gemini key was safely stored took some trial and error.
Cross-platform development: Balancing frontend responsiveness with backend reliability was tricky—especially when switching development machines.
Accomplishments that we're proud of Created an end-to-end working AI pipeline between frontend input and backend Gemini summary generation.
Designed an emotionally supportive tool that is both functional and compassionate.
Built an intuitive UI that's easy to use for both children and caretakers.
Learned and implemented FastAPI and Google’s Gemini API from scratch within the short hackathon window.
What we learned How to build full-stack apps using FastAPI and React.
The ins and outs of integrating generative AI (Gemini) into real-time applications.
How to write empathetic prompts to generate accessible language suited for users with cognitive or emotional challenges.
The importance of accessibility-first design for inclusive apps.
What's next for MindMirror Storing mood history over time using a real database to track emotional trends and insights.
Visual dashboards for parents to understand patterns and triggers in their child’s emotional state.
Voice-based input/output for users who may struggle with typing or reading.
Integration with therapists or educators, allowing shared emotional logs and better-informed care.
Mobile-friendly version to allow on-the-go check-ins anytime, anywhere.
Built With
- fastapi
- gemini
- python
- tailwind
- typescript
Log in or sign up for Devpost to join the conversation.