Voice agents that listen, act, and reply
Mistral’s Voxtral 3 combines speech recognition, reasoning, tool calls, and spoken responses in one streaming system. Its voice agent SDK can handle interruptions and complete tasks during live phone or web conversations, such as checking a bill or booking an appointment.
The family includes transcription across more than 50 languages and speech generation across more than 35 languages and 50 dialects. It is available through Mistral’s API, with self-deployment offered for organizations that need to keep audio on their own infrastructure.
Gemini brings personal context into the car
Google is rolling out Gemini Personal Intelligence on Android Auto. With permission, it can draw from Gmail, Calendar, Photos, Maps, YouTube, and Search while you drive.
That enables requests such as finding a flight gate and checking its status, navigating to an appointment from calendar details, or locating a place from a photo. App connections are opt-in and can be disabled; the rollout starts now on Android Auto and is due on cars with Google built-in in the coming months.
Also
- ElevenLabs simplifies realtime voice agents Its new Mini and Max speech-to-speech models support more than 90 languages. Existing OpenAI Realtime API integrations using WebSocket can switch by changing the endpoint and model name, while WebRTC integrations must move to WebSocket first.
- Gemini adds a small-business workspace The new hub connects services including Google Workspace, Canva, Shopify, and Xero, while retaining uploaded brand styles for future visuals.
- Put agent work on one dashboard Ben Tossell describes an all-in-one site for viewing agent activity, files, email, customer records, domains, and tasks, with a linked walkthrough and copyable instructions.
Pass it on