ShipVoice
The voice AI platform you own.
Every client, one control plane.
Run a voice agent for every client, on LiveKit Cloud or your own servers.
Be first when it opens. No spam.
- LiveKit Cloud and Inference
- Any provider
- Billed through your Stripe
- 0% platform fee
The ShipVoice console, shown with demo data.
Every layer is a plugin. Swap anything. Rewrite nothing.
- Transport
LiveKit Cloud or your own LiveKit.
- Speech
Deepgram or any speech model.
- Model
OpenAI or any LLM, direct or through LiveKit Inference.
- Voice
Cartesia or ElevenLabs.
- Carrier
Twilio or Telnyx.
- Pattern
Ten ways to run a call.
Bring your own keys.
Client to live agent. Three steps. Every time.
- Step 1
Add the client.
Their own space. Nothing leaks.
- Step 2
Pick a pattern.
Ten proven ways to run a call.
- Step 3
Go live.
Every call, transcript and cost. Billed with your margin.
Ten call patterns. Fast and right, as agents grow.
One long prompt slowly breaks. Patterns don't.
Read the pattern guide-
Solo
One agent holds the whole call with one prompt and every tool.
Pattern 01 -
Supervisor
One agent keeps the call and lends it to focused tasks, each of which hands back a typed result.
Pattern 02 -
Relay
Control passes from agent to agent. Each owns a phase with its own voice, tools and permissions.
Pattern 03 -
Guided Path
The flow fixes the order of steps. The caller can step back to fix an earlier answer.
Pattern 04 -
Observer
A silent second model reads the transcript and steers the speaking agent. It never talks to the caller.
Pattern 05 -
Talker and Thinker
A fast model answers every turn while a slower reasoning model plans alongside it.
Pattern 06 -
Background Worker
A long job leaves the turn. The agent keeps talking, and the result rejoins the call, or reaches the caller by text.
Pattern 07 -
Scatter and Gather
One question goes to several workers at once. Their answers are merged into a single reply.
Pattern 08 -
Bookends
Work that does not need the caller moves before the first word and after the last.
Pattern 09 -
Warm Handover
The agent brings in a person, briefs them privately, then steps out.
Pattern 10
The client console. Their calls. Only theirs.
Client logins and roles coming soon
- Calls and transcripts
- Spend, with your margin
- Roles for your team
Rent a platform, or own one. Own it, and the margin is yours.
| Feature |
Managed
Vapi, Retell, Bland |
ShipVoice
you own it |
|---|---|---|
| Every client separated | ||
| Phone numbers and SIP | ||
| Swap any provider | ||
| 0% platform fee | ||
| You own the code | ||
| Runs on your own accounts | ||
| Cost per minute | $0.15+ | provider cost |
Managed platforms run $0.15 to $0.36 a minute, before your markup. Work out your margin or see the full comparison
What's inside. Open kernel. Agency edition.
Open kernel
- Plugin slots
Every layer swappable.
- LiveKit Agents
A proven realtime runtime.
- Checked at boot
Bad combinations fail before a call does.
Agency edition
- Control plane
Every client, one install.
- Ten call patterns
Ready to compose.
- Client console
Calls, transcripts and spend.
- Per-minute billing
Through your Stripe, with your margin.
Voice core and billing run today. The open kernel and agency layer ship at launch.
New to voice AI? Start with the free Voice AI Primer
I build voice AI infrastructure. ShipVoice is the platform. The agent on this page runs on it.
Questions. Answered.
Who is it for?
Agencies that build and run voice agents for clients. Clinics, trades, restaurants. One install, every client separated.
What is free and what is paid?
The kernel will be open source and free. The agency edition adds the control plane, client console, billing, call patterns and premium plugins.
Can my clients log in?
Soon. Each client gets a read-only view of their own calls, transcripts and spend.
How does billing work?
Every minute is metered. You set the margin. Clients pay through your Stripe.
Where does it run?
Your choice. LiveKit Cloud for calls and phone lines with the rest on Railway, all on LiveKit Cloud, or fully on your own servers. ShipVoice hosts nothing.
Does it work with LiveKit Inference?
Yes. Pick a model by name and pay through your LiveKit account, or connect each vendor with its own key.
How do I get access?
Join the waitlist. One email when it opens.
Own your voice AI platform.
Early access for agencies.
Be first when it opens. No spam.