Voice first
Real conversation, not voice commands. Interrupt it mid-sentence, walk away from the screen, pick up on another device. It keeps talking even while it updates itself.
Talk to an assistant
that answers to you.
VoiceAssist is a voice-first personal assistant that runs your errands, your calendar and your code, on servers you choose. Your memory and model keys stay on your assistant’s server. In hosted mode, its operator can technically access them.
Get VoiceAssistBuilt for hands-free, built for your stuff
Most assistants are walled gardens. VoiceAssist is designed to reach into the machines and tools you already own.
Bring your own servers
Point it at your workstation, homelab or cloud box. Coding agents work there over SSH, in your workspaces, with your tools, while you talk them through it.
Plug in any MCP tool
Robots, VR rigs, home automation, your own services: anything that speaks the Model Context Protocol becomes something you can ask for out loud.
Asks before it acts
Every action passes a permission check. When something needs your say-so, it asks you, out loud if you're on a call, and remembers what you decided.
Lives on your phone
Alarms that ring offline, calendar, navigation, texts, and Health Connect data, all read on demand from your device. No background location tracking.
Open at the core
The parts that hold your data are open source, so you can read them, run them yourself, or pay us to run the heavy lifting for you.
One private assistant, together
Your brain and voice server run together in your own isolated virtual machine. Shared services handle invitations, sign-in and access.
- Your deviceSign in onceUse an invitation to create your account, then your username and password.
- Shared accessYour accountConnects you to your assigned assistant. Other users cannot access it.
- Your private VMBrain + voiceYour conversations, settings, tools and model credentials live together.
- Your choiceModels and toolsConnect only the providers and services you choose.
Privacy, with clear limits
Separate user machines
Each hosted user gets a private VM. Hosting operators can technically access its data; VM isolation is not a promise that operators cannot read it.
Your connection
The private SSH pilot uses a trusted browser proxy that can read application traffic.
Public assistant connections use TLS directly to your assigned VM. The shared entry service routes encrypted assistant traffic.
Your model provider
The selected model receives your requests and context. Its retention and training terms depend on the provider account and feature. We do not promise zero retention.
Credentials and history
Model keys and conversation history stay on your VM, outside the shared account database. Invited alpha users will receive separate provider keys; operator test accounts may temporarily use the operator’s keys.
Voice and transcripts
We do not record microphone audio. Transcripts and voice context may persist on your VM under its retention settings.
No tracking on this site
No analytics, ads, third-party fonts or trackers. We plan to fund VoiceAssist through its users.
A small, free, invite-only alpha
Four users, including the founder. No payment or subscription is required for this pilot. Paid plans come later.
Use your invitationQuestions
Do I need my own server?
No. In this alpha, we provision a separate VM running your brain and voice server together.
Which AI models does it use?
VoiceAssist supports models from providers such as OpenAI and Google. Availability depends on the keys configured on your VM. Provider terms and charges apply.
Can it really change things on my machines?
Only the ones you connect, only inside the workspaces you choose, and only after passing a permission check. Anything risky is put to you first.
What if I cancel?
Contact support for an identity-checked, encrypted export or account deletion. These operations are operator-assisted during alpha. The privacy notice explains backup retention and deletion confirmation.
Why no investors?
Investors need a return, and in this market that usually means your data. Being paid directly by the people who use it keeps our goals the same as yours.