From Pymnts · PYMNTS · · 1m 10s
Google pushes Gemini deeper into voice AI
Google’s new Gemini voice models aim to make enterprise AI agents cheaper, smarter and easier to talk to.
In brief
Google’s new Gemini voice models aim to make enterprise AI agents cheaper, smarter and easier to talk to. Voice AI is moving from simple chat toward agent systems that can reason, act and handle business workflows. Originally reported by Pymnts.
Two new models
Google is rolling out Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking to power voice agents for developers and enterprises.
Built for different jobs
Gemini 3.8 Live targets scale and cost efficiency. Extended Thinking is meant for harder tasks, with pricing Google says can compete with other frontier models.
More natural talk
Both models add near real-time reasoning so voice agents can respond faster and feel less scripted during conversations.
“make conversing with AI feel more intuitive and intelligent”
Enterprise voice stack
Google is pitching the models as production-ready building blocks for companies that want AI agents in customer support, sales and internal workflows.
Voice moves to action
PYMNTS frames voice as a bridge between large language models and transactions, because speaking can reduce friction when customers are ready to buy.
Workspace gets louder
Google is also adding voice AI to Gmail, Docs and Keep, letting users search emails, build documents and turn spoken ideas into lists or notes.
“using just your voice”
The takeaway
Voice AI is moving from simple chat toward agent systems that can reason, act and handle business workflows.
