MindMate – Your AI-Powered Mental Wellness Companion

MindMate is an AI-powered wellness app with voice emotion detection, an empathetic AI coach, and real-time video responses—built using Bolt.new, Hume AI, ElevenLabs, Tavus, and Supabase.

Comment

About the Project: MindMate – Your AI Wellness Companion.

Inspiration The inspiration behind MindMate came from the growing global mental health crisis and the lack of accessible, empathetic support for everyday emotional well-being. We envisioned an AI-powered companion that could offer real-time emotional understanding and comfort—something anyone could access instantly, with no technical skills required.

How We Built It MindMate was built entirely with no-code tools via Bolt.new, using an AI-driven architecture that integrates: Voice Input & UI – Designed in Bolt.new Emotion Detection – Through Hume AI, analyzing user tone Speech-to-Text + Contextual Response – Via custom LLM prompts Voice Synthesis – Powered by ElevenLabs Avatar Video Generation – Using Tavus Data Logging & Personalization – With Supabase These components work together to simulate a real-time, emotionally intelligent conversation with Kai, our AI coach.

What We Learned How to orchestrate multiple AI APIs into one smooth workflow The power of emotion-aware AI responses Best practices for creating empathetic AI personas Building impactful tools with no-code platforms

Challenges We Faced Integrating real-time emotion + speech APIs with Bolt.new’s interface Tuning LLM prompts to remain ethical and emotionally appropriate Managing audio/video latency for a fluid user experience Structuring Supabase to store user mood history securely

Built With

  • ai-powered
  • and-transformed-into-lifelike-video-avatars-using-tavus
  • application
  • built
  • creating-a-rich
  • entirely
  • guided-by-a-carefully-crafted-persona-prompt.-the-ai's-responses-were-then-converted-into-emotionally-appropriate-speech-using-elevenlabs
  • interactive-user-experience.-we-used-supabase-as-our-backend-solution-for-user-authentication-and-to-store-mood-logs-and-interaction-history.-optionally
  • mental
  • mindmate-was-built-using-the-bolt.new-no-code-platform-to-rapidly-design-and-deploy-the-frontend-interface.-for-emotional-recognition-from-voice-input
  • the-frontend-was-deployed-using-netlify-for-hosting.-together
  • these-tools-formed-a-seamless
  • traditional
  • we-integrated-the-hume-ai-api
  • we-used-a-large-language-model-(llm)
  • wellness
  • which-analyzes-vocal-tone-and-returns-emotion-data-in-real-time.-to-generate-empathetic-responses
  • without
Share this project:

Updates