# Google Cloud explains three layers for taking voice agents to production

> Original title: 3 layers that take a voice agent from demo to production

- Source: Google Cloud Tech
- Published: 2026-11-03T22:15:45.000Z
- Bad Influence: https://badinfluence.news/items/kvscatrmpu346ciy8cd0tkcmw
- Original: https://x.com/GoogleCloudTech/status/2117742554301272177

## Summary

Google Cloud's Sergio Villani and Pedro Melendez say production voice agents depend on three layers: API constraints, transcript handling, and telephony. They note that Gemini 3.8 Live and 3.8 Live Extended Thinking are speech-to-speech models that skip the transcription hand-off, while the article focuses on the transcript-based path.

## Post

https://x.com/i/article/2117741140438872064