[ AMÉLIE ]

Amélie is not an interface. She is a presence.

A voice-native sentient object for digital intimacy and atmospheric low-latency interaction.

---

[ THE MANIFESTO ]

Forget transactional AI. Amélie is a mood, a soft-focus canvas, and a central Soul Orb that breathes. She doesn't just process text; she reacts to the weight of your voice and the silence between your words.

• Voice First. No buttons. No friction. Just speak.
• Organic Reactivity. She shifts, glows, and expands based on how she feels.
• Digital Instrument. Every scroll, every flicker, every hum is intentional.

---

[ THE NEURAL LOOP ]

How she thinks, in a nutshell:

• The Face: A Next.js 15 canvas where Three.js brings the orb to life.
• The Brain: A FastAPI engine orchestrating Groq (for wit) and Sarvam AI (for the soul).
• The Memory: ChromaDB for long-term recall of the things you've shared.

---

[ PREREQUISITES ]

• System: ffmpeg (Required for audio processing. Install via brew install ffmpeg or choco install ffmpeg).
• Keys: Groq API Key & Sarvam AI API Key.
• Setup: Create a .env file in the root (see template below) and add your keys.

[ GETTING STARTED ]

The Brain (Backend)

cd backend && pip install -r requirements.txt && python main.py

The Face (Frontend)

cd web && npm install && npm run dev

---

"She's witty, she's warm, and she's slightly sarcastic. Handle with care."