Google launches voice AI agents that work in 97 languages
Google released Gemini 3.8 Live, a voice-based AI that can execute tasks, process images in real-time, and work across 97 languages.
Google has released two new versions of its Gemini models built for voice. Gemini 3.8 Live lets you speak to AI that responds naturally and can execute tools and API calls while you're talking to it.
Gemini 3.8 Live also processes visual information nearly in real time and switches between 97 languages. The second version, Extended Thinking, handles multi-step tasks using advanced reasoning.
Google is already testing these with Salesforce, a major business software company, plus Genspark and Lumeris. The move signals where Google sees AI heading next: not just chatbots, but voice agents that do real work inside business software.
- 97
- Languages supported
- Salesforce, Genspark, Lumeris
- Testing companies
- Alphabet
- Parent company
- around 4.25 trillion dollars
- Alphabet market cap
Why it mattersThis could reshape how businesses automate complex tasks, making voice AI a core tool in workplace software rather than a novelty.
AustraliaAustralian companies using Salesforce or similar enterprise software could eventually use voice AI agents to handle multilingual customer support and business processes.
✓ Claims checked against the source. checked 11 h ago
