MirrorMind (Talk to yourself. Know yourself)

An AI-powered mirror of your personality — that talks back in your voice.

Comment

💡 Inspiration

We often wonder how others perceive us — our tone, our reactions, our presence in conversations. But what if we could experience ourselves the way others do?

That’s the idea behind MirrorMind: a tool for self-reflection powered by AI, voice cloning, and personality analysis.

When we discovered Bolt.new, it gave us the perfect foundation to bring this vision to life. From project setup to voice-AI integrations, Bolt.new made complex architecture feel intuitive — so we could focus on the emotion, not the boilerplate.

🧠 What it does

MirrorMind allows users to:

  • Take a quick psychological quiz to assess their current personality traits.
  • Upload a voice sample (or use a default one).
  • Chat with an AI version of themselves — shaped by their own traits.
  • Hear responses spoken back in their own voice, with tone and attitude aligned to their weakest trait (if any).

It’s a voice-powered mirror of “you right now” — raw, confronting, and eye-opening.

🛠 How we built it

This entire project was initialized on Bolt.new, which handled:

  • Project scaffolding (React + Vite + Tailwind).
  • Secure OpenAI and ElevenLabs integrations via Supabase Edge Functions.
  • Authentication setup and routing.

We then expanded it with:

  • A custom quiz engine using Context API to track user traits.
  • Dynamic rendering of the radar chart for results visualization.
  • Interactive chat interface where AI leads the conversation based on personality weaknesses.
  • Voice cloning and TTS with ElevenLabs (based on Bolt's examples).
  • Secure backend calls via Supabase, avoiding any client-side key exposure.

Bolt.new’s examples and documentation gave us the confidence to push boundaries in multi-modal interaction and emotional UX — without getting stuck in low-level setup.

🧩 Challenges we ran into

  • Managing async voice cloning and playback smoothly across browsers.
  • Designing emotionally intelligent AI prompts to reflect traits in a believable way.
  • Persisting user state (traits, voice, chat context) without bloating the frontend.
  • Balancing performance and emotional depth within a short hackathon window.

🏆 Accomplishments that we're proud of

  • A fully functional prototype combining OpenAI + ElevenLabs + Supabase — all in one flow.
  • A chat experience that feels personal and confronting, yet empathetic.
  • A voice interaction that lets people hear how they emotionally affect others.
  • Staying true to the goal: building something deeply human with the power of AI.

📚 What we learned

  • Building emotional AI is about tone and delivery, not just raw intelligence.
  • Voice adds a layer of reality that text alone cannot match.
  • Bolt.new dramatically speeds up secure, real-world AI app development — without sacrificing flexibility.
  • The best user experiences often come from emotion-first, tech-second thinking.

🚀 What's next for MirrorMind (Talk to yourself. Know yourself)

We plan to explore:

  • Multiple AI personalities (e.g. “your younger self”, “your confident self”, etc.).
  • Long-term journaling and personality tracking.
  • Voice tone modulation based on emotional context.
  • A toolset for coaches and therapists to help clients hear how they come across.

Ultimately, MirrorMind aims to become a mirror for your soul — not just your voice.

Built With

  • bolt.new
  • context-api
  • elevenlabs-api
  • framer-motion
  • lucide-icon
  • openai-api
  • react
  • supabase
  • supabase-edge-functions
  • tailwind-css
  • typescript
  • vite
Share this project:

Updates