Mind Mate - AI

Emotion-aware voice AI for human-centric support.

Comment

Inspiration

Mental health support is hard to access when people need it the most — late at night, in private moments, or when they don’t feel ready to talk to another person. Mind Mate – AI started from a simple idea: give users a safe, non-judgmental space to talk, anytime. The goal wasn’t to replace professionals, but to act as a first step — someone who listens.

What it does

Mind Mate – AI is a conversational AI companion designed to support users emotionally. It allows users to:

Chat freely about their thoughts and feelings

Receive calm, empathetic responses

Feel heard without judgment or pressure

Get encouraged toward professional help when needed

It intentionally avoids medical diagnosis or therapy claims.

How we built it

Frontend: Deployed on Netlify with a clean chat UI

Backend: FastAPI hosted on Railway

AI: Google Gemini (Generative AI) for responses

Architecture: Modular services (chat model, Gemini service, Firestore service)

Security: Environment variables for API keys, CORS-controlled API access

Deployment: Dockerized backend for consistent builds

The system sends user input to the backend API, processes it through Gemini with a structured prompt, and returns a response to the UI.

Challenges we ran into

Backend deployment issues (Railway + environment variables)

Import errors and service-layer wiring (generate_ai_reply)

Handling API failures gracefully without breaking the UI

CORS configuration between Netlify and Railway

Ensuring responses stayed empathetic but non-medical

Debugging under time pressure was the hardest part.

Accomplishments that we're proud of

Fully deployed end-to-end system (frontend + backend + AI)

Clean service-based backend architecture

Graceful error handling instead of app crashes

A UI that feels calm and human, not robotic

Finishing despite deployment and integration issues

What we learned

Real deployment problems teach more than tutorials

Small import or config mistakes can break the whole system

AI apps need strong prompt control, not just API calls

CORS, env vars, and deployment matter as much as code

Building responsibly for mental health is critical

What's next for Mind Mate – AI

Conversation memory and context handling

Mood tracking over time

Multilingual support

Better fallback responses when AI fails

Optional escalation to verified mental health resources

Mind Mate – AI is a foundation, not a finished product.

Built With

Share this project:

Updates