Voice-first form intake for businesses.
Upload a blank fillable PDF, generate a shareable interview link, and let clients complete forms through a guided AI conversation.

auraforming.ai turns complex paperwork into a structured, multilingual voice workflow.
The business uploads a PDF once; the platform extracts fields, creates an agent_id, and serves a client-facing interview page. During the interview, ElevenLabs handles speech I/O while Gemini evaluates each answer field-by-field and drives the next prompt. At completion, the backend writes answers into the PDF and stores the intake for admin review.

• Video demo: [TO BE ADDED]()

• Frontend: React, Vite, React Router
• Backend: Flask, Flask-CORS
• Data: SQLite + filesystem storage
• PDF: PyMuPDF (fitz)
• AI reasoning: Gemini API (gemini-2.0-flash by default)
• Voice: ElevenLabs STT + TTS

python3 -m venv .venv
source .venv/bin/activate
pip install -r backend/requirements.txt
npm --prefix frontend install

Set required env vars:

export GEMINI_API_KEY="your_google_ai_key"
export ELEVENLABS_API_KEY="your_elevenlabs_key"
export ELEVENLABS_VOICE_ID="your_voice_id"

Run:

python backend/app.py
npm --prefix frontend run dev

• Backend: http://127.0.0.1:5050
• Frontend: http://127.0.0.1:5173

• Setup and startup: docs/SETUP_AND_RUN.md
• System architecture and API map: docs/SYSTEM_OVERVIEW.md
• Docs index: docs/README.md
• Gemini details: docs/GeminiIntegration.md
• ElevenLabs integration notes (no Agent mode): docs/ElevenlabsAgentConfig.md