# VoxPolicy

- **Event:** [Built with Opus 4.6: a Claude Code hackathon](https://cerebralvalley.ai/e/claude-code-hackathon)
- **When:** Feb 10 at 12:00 PM – Feb 17 at 10:00 AM (EST)
- **Where:** Location TBA
- **Team:** [Udam Liyanage](https://cerebralvalley.ai/u/udamliyanage)
- **GitHub:** https://github.com/UdamLiyanage/vox-policy
- **Demo video:** https://drive.google.com/file/d/1PU6-DTTwOMBFVrq1sZA2g0qpC32Vj2QP/view
- **Gallery:** https://cerebralvalley.ai/e/claude-code-hackathon/hackathon/gallery
- **Page:** https://cerebralvalley.ai/e/claude-code-hackathon/hackathon/gallery/15

VoxPolicy is a reasoning-first multilingual voice agent for call centers showcasing Claude's transparent reasoning, autonomous tool use, and cross-lingual understanding.

Key Features:
- Automatic customer identification by phone number
- Real-time voice (Deepgram STT + ElevenLabs TTS)
- Visible AI reasoning — Claude's thinking displayed live
- Multilingual — English, Spanish, Mandarin with auto-detection
- Customer memory persists across interactions
- Autonomous tool use — search policies, update memory, or escalate
- Smart escalation with structured handoff summaries

Tech Stack: React, FastAPI, Claude Sonnet with extended thinking, Deepgram, ElevenLabs, SQLite

Why "Amplify human judgment": VoxPolicy doesn't replace agents — it makes them better. Complex situations escalate with full context: customer history, sentiment, churn risk, and recommended actions. Visible reasoning lets supervisors understand every AI decision, building trust in autonomous customer service.

---

Markdown version of https://cerebralvalley.ai/e/claude-code-hackathon/hackathon/gallery/15. Site index for agents: https://cerebralvalley.ai/llms.txt · full text: https://cerebralvalley.ai/llms-full.txt