Voice-control coding agents
from anywhere
Turn your commute into coding time: start and steer Claude Code and Codex sessions running on your own machine, over SSH, by voice, while you drive. Or with your phone in your pocket and earbuds in, on a walk, or away from the desk. Use it from the phone app, your car (CarPlay / Android Auto), or your watch. Works with any LLM, including Claude, GPT, and Gemini.
Built for your ears, not your eyes: agent output is spoken and summarized, approvals are verbal, and the screen is optional by design.
The self-hosted backend is free — run it on your own machine. Or skip the setup with DashVox Cloud Ultimate: a fully managed VM with pre-configured coding agents, ready to go.
Agent via SSH:
frontend — running tests, 3 passing
backend — awaiting approval on migration
tests — completed, all green
See it in action
DashVox, hands-free
Reclaim your commute
You could be coding right now
Most developers lose close to an hour a day to the commute. DashVox turns that drive into coding time: review diffs, approve steps, and steer Claude Code or Codex sessions while you're stuck in traffic — reclaiming time you'd otherwise lose. Code while driving, fully hands-free with your phone's voice-driven mode, or from the native CarPlay app's simple controls.
Fully hands-free mode
Put your phone on the dash and use the voice-driven driving mode — speak your request, hear the reply, no taps. No CarPlay needed, so it works in any car.
Native CarPlay controls
Prefer CarPlay? Open a chat and use simple controls — press the main button to record, press to send, back to cancel. Glanceable, the way CarPlay is meant to be used.
Spoken output, eyes up
Either way, agent output is read aloud and summarized — so you keep your eyes on the road.
DashVox is voice-first and designed to minimize distraction while driving, but you are always responsible for driving safely and following local law.
Capabilities
Everything you need,
nothing you don't
Voice-First AI Chat
Talk to any LLM (e.g. Claude, GPT, Gemini) right on your phone — hands-free while you drive, from CarPlay, or on your watch. Responses read aloud, input via speech.
Remote Session Management
SSH into your dev machine from wherever you are. Start, stop, and switch AI coding sessions. Approve prompts and check output by voice.
Agent Mode
Your AI autonomously manages multiple coding agent sessions (e.g. Claude Code, Codex). Tell it to spin up frontend, backend, and test sessions — it handles everything.
Web Search Integration
AI can search the web on your behalf and summarize results. Stay informed without touching a screen.
Smart Connectivity
Exponential backoff, background ping, auto-resume. Idempotent message delivery means no duplicates after dead zones.
Security First
Self-host and no account is needed — no Firebase, your code and data never leave your machine. Cloud mode adds Firebase Auth sign-in and stores your SSH keys server-side to connect on your behalf, with at-rest encryption planned.
Workflow
Three steps to AI on your commute
Open the app — phone, car, or wearable
Use DashVox on your phone, or in the car: a fully hands-free, voice-driven driving mode (phone on the dash) plus native CarPlay and Android Auto apps with simple controls. Voice replies reach Apple Watch and Wear OS too — with smart glasses (Ray-Ban Meta) in exploration.
Speak your request
Voice is transcribed in real-time. Ask questions, start coding sessions, or search the web.
AI responds aloud
Responses are spoken via TTS. Review summaries on the display, and approve actions by voice or with a single CarPlay control.
Built with
Why DashVox
How DashVox compares
Other tools solve a piece of this. DashVox is built specifically to drive real coding agents, on real machines, by voice — from the car.
Sandboxed voice coding tools
Run your prompt in a throwaway cloud sandbox that resets — no access to your real repo, your branches, or your running services.
DashVox
Connects to your real machine over SSH (or a managed VM you control). Your actual repo, your branches, your environment.
Screen-bound SSH / terminal apps
Give you a tiny terminal you have to read and type into — unusable while driving.
DashVox
Voice-first: agent output is spoken and summarized. Drive it fully hands-free by voice, or with a couple of glanceable CarPlay controls — no tiny terminal to read or type into.
Desktop-tethered remote controls
Need a desktop left running a screen-share or remote session for you to drive.
DashVox
Talks to the coding agent over SSH directly — or to DashVox Cloud Ultimate — with no desktop babysitting a session.
Platform
Available everywhere you go
iOS / Apple Watch
Native SwiftUI app with CarPlay support. Voice reply from Apple Watch via mirrored notifications.
Download on the App StoreAndroid / Android Auto
Android Auto ready. Voice reply also works from Wear OS watches via connected-device notifications. Ray-Ban Meta glasses support is in exploration (Meta API permitting).
Coming soonSelf-Hosted Server
Run the backend on your own machine. Java 21, cross-platform. All data stays local.
Learn moreFAQ
Frequently asked questions
Can I use Claude Code while driving?
Yes. Put your phone on the dash and use the fully hands-free, voice-driven driving mode — speak your request, hear the reply, no taps. Or run it from the native CarPlay app with simple press-to-talk controls (press to record, press to send, back to cancel). Either way the agent runs on your own machine over SSH — or on a managed DashVox Cloud VM — and reads its output back to you aloud.
How do I run coding agents from CarPlay?
Install the DashVox iOS app, connect your dev machine under Settings → SSH Profiles, then open DashVox from the CarPlay display. It's a native CarPlay app with simple controls: open a chat, press the main button to record your request, press again to send (or back to cancel), and the agent's replies are read aloud. Prefer not to use CarPlay? The phone's hands-free driving mode does the same thing fully by voice.
Does it work on Apple Watch?
Yes. You get spoken session updates and can reply by voice from Apple Watch (and Wear OS on Android) via mirrored notifications — useful when you step away from the car too.
Do I need my own machine to use it?
No. Self-hosted mode runs on your own hardware for free, but if you would rather not run anything yourself, DashVox Cloud Ultimate gives you a managed VM with pre-configured coding agents, ready to go.
Is it safe to code while driving?
DashVox is designed to keep your eyes on the road. The fully hands-free driving mode works by voice alone — like taking a phone call — while the CarPlay app keeps interaction to a few large, glanceable controls, the way CarPlay is meant to be used. Agent output is always read aloud. You are still responsible for driving safely and following local law.
Do I have to look at the screen to use it?
No. DashVox is built ears-and-speech-only: agent output is read aloud and summarized, and you approve actions by voice. The screen is there when you want it, never required.
Which AI models does DashVox work with?
Any major LLM — Claude, GPT, and Gemini — plus custom endpoints for self-hosted or enterprise models. You can switch models mid-conversation.
Can my company use DashVox?
Yes. DashVox Enterprise gives teams managed cloud machines, org-level seats and budgets, SSO, and admin controls. Contact sales on the Enterprise page.
DashVox is on the App Store
Download DashVox for iOS now. Android is approved with a public beta on the way. Both unlock the voice-first experience on your phone, car, watch, and glasses.