ShipShip Voice Agents that makes money , not demos
The complete FastAPI + React + LiveKit Agents SaaS template with 10+ ready-to-deploy voice agents. Optimized from internals using OpenRTC and profiled every details with VoiceGateway. Launch a profitable voice AI business in days, not months.
Be first when it opens. No spam.
The agent you just talked to was built with ShipVoice. Open-source core: livekit-starter ↗
Swap any provider · bring your own keys
OpenAI 
Cartesia 
Twilio 
Built on LiveKit, the realtime stack behind OpenAI, Character.ai, and Retell.
From idea to shipped, in three steps.
Speed is the point. The AI engineer does the scaffolding, you review and ship.
-
Open it in your coding agent
Claude Code, Cursor, whatever you use. The repo ships with the subagents and the ShipVoice pattern baked in.
-
Describe your agent in a line
The built-in AI engineer (architect, prompt engineer, task scaffolder, data and keyterms generator, reviewer) scaffolds it on the pattern.
-
Run it and ship
A live agent, plus auth, a call dashboard, per-minute Stripe billing, and one-command deploy already wired in.
The plumbing you skip.
An agent talks in an afternoon. The business around it (billing, auth, dashboard, deploy) is where the weeks go.
- Per-minute metering (STT · LLM · TTS · SIP) ~20h
- Telephony: SIP / PSTN + 10DLC scaffolding ~15h
- Auth, accounts, and a call dashboard ~20h
- Usage → Stripe metered billing + webhooks ~15h
- Deploy, ops, and scaling ~10h
- 1 git clone templates + the AI engineer
- 2 describe your agent subagents scaffold it on the pattern
- 3 ship billing, auth, and deploy already wired
Start from a working receptionist, not a blank file. Estimate your per-minute margin →
Own it, don't rent it.
Own the code, the agent, and the margin, instead of renting a black box per minute.
|
Free starter
livekit-starter |
Managed
Vapi · Retell · Bland |
ShipVoice
own it | |
|---|---|---|---|
| Per-minute metering | ✕ | ✓ | ✓ |
| Telephony / SIP | ✓ | ✓ | ✓ |
| Call dashboard | ✕ | ✓ | ✓ |
| You bill YOUR users | ✕ | ✕ | ✓ |
| You own the code + margin | ✓ | ✕ | ✓ |
| All-in cost / min | provider cost | $0.15+ | provider cost |
| can't charge | rented forever | a product you sell |
Managed platforms land at $0.15 to $0.36 all-in per minute. Above ~20,000 minutes a month, owning it wins.
See the full comparison →
Everything in the LiveKit SaaS boilerplate
A voice-AI boilerplate in three layers: the voice core, the money layer that lets you charge, and the pieces that help you sell.
Voice core
- LiveKit Agents pipeline -20h
STT, LLM, TTS, turn-taking, swappable providers
- Telephony, in + outbound -30h
SIP and PSTN, real phone calls
- Browser + mobile widget -15h
Embeddable mic for web and mobile
Money layer
- Stripe per-minute metering -40h
Meter each minute, add your margin
- Auth + per-user accounts -25h
Sign-up, login, per-user usage
- Call dashboard + costs -25h
Recordings, transcripts, per-call cost
Ship-it layer
- AI engineer built in -40h
Subagents scaffold a new agent from one line
- shipvoice init + 10 templates -30h
Pick a template, add keys, scaffold, run
- One-command deploy -20h
Docker + Fly / Render / LiveKit Cloud
About 160 hours of undifferentiated plumbing, done. New to building voice agents? Read the free Voice AI Primer.
I'm Mahimai. I build voice agents; the one on this page is mine, running on ShipVoice. I got tired of rebuilding auth, per-minute metering, billing, and telephony for every voice product, so I packaged it. The core is open source (livekit-starter); ShipVoice is that core plus an AI engineer and templates, so building and charging for a voice product takes an afternoon, not a quarter.
Questions
Do I need to be a voice-AI expert? +
No. ShipVoice ships an AI engineer: a set of subagents that scaffold a working agent on a proven pattern from a one-line description. Open the repo in any coding agent (Claude Code, Cursor, whatever you use), say what you want, and it builds on the ShipVoice structure. You review and ship.
Is this just the free LiveKit starter? +
No. The free starter (our open-source livekit-starter) gets you a working agent. ShipVoice adds the speed and the business: the AI engineer, 10 templates, a shipvoice init wizard, plus per-minute Stripe billing, auth, a call dashboard, and telephony. The part that takes you from idea to a monetized product in an afternoon.
How do I get access? +
Join the waitlist and you will be the first to know when ShipVoice opens. Drop your email below; no spam, just one note when it is ready.
What is the stack? +
LiveKit Agents (Python and TypeScript) for the voice backend, a Next.js dashboard and frontend, Stripe for metered billing, Postgres for data, and Docker for deploy. Model providers are swappable.
What does the license cover? +
A single-developer license with unlimited projects and lifetime updates. A team license comes later.
Do I still have to handle telephony registration? +
Telephony is wired in, but 10DLC and A2P carrier registration is an external step (roughly two weeks at the carrier) that no code can remove. ShipVoice ships the scaffolding and a guide so you are not figuring it out alone.
What if LiveKit ships billing themselves? +
You still own the code and keep your margin. That is the one thing a platform will never hand you, and it is the reason to own your stack instead of renting it.
Be first in line.
Join the waitlist and you will be the first to know when ShipVoice opens.
Be first when it opens. No spam.