Quickstart
Get your first AI voice agent running in minutes.
Self-host
Deploy Sniko on your own infrastructure.
Agents
Create and configure conversational voice agents.
Integrations
Connect ElevenLabs, Twilio, calendars, and payment providers.
Who Sniko is for
Sniko is designed for businesses and developers who want to deploy conversational AI agents over the phone without building infrastructure from scratch. Common use cases include:- Customer support — Route callers to an AI agent that answers FAQs from a knowledge base.
- Lead qualification — Run outbound call campaigns that ask screening questions and log structured responses.
- Appointment booking — Let agents schedule meetings and sync directly with Google and Microsoft Calendar.
- KYC and surveys — Collect identity verification data or survey responses through guided voice conversations.
- Sales outreach — Automate first-contact outbound calls at scale using CSV contact lists.
Core integrations
ElevenLabs
Powers all voice synthesis, conversational AI agents, and knowledge base retrieval. Every agent in Sniko maps to an ElevenLabs Conversational AI agent.
Twilio
Provides inbound and outbound telephony. Import Twilio phone numbers, run batch calling campaigns, and handle SMS messaging.
Payments
Stripe, Razorpay, and Paystack are supported for subscription billing. Admins define pricing plans that users subscribe to.
Architecture overview
Sniko is a monolithic Laravel 12 application with a Vue 3 SPA frontend served via Inertia.js. There is no separate API — the frontend communicates directly with the Laravel backend through Inertia page props and form submissions.| Layer | Technology |
|---|---|
| Backend | Laravel 12 (PHP 8.2+) |
| Frontend | Vue 3 + Inertia.js v2 |
| Styling | Tailwind CSS v4 |
| Database | SQLite (default) or MySQL |
| Queue | Laravel database queue driver |
| Voice AI | ElevenLabs Conversational AI |
| Telephony | Twilio Voice + SMS SDKs |
| Billing | Stripe, Razorpay, Paystack |
| MCP server | Node.js (PM2, port 3001) |
Request flow
- The user’s browser loads a Vue 3 page through an Inertia response from Laravel.
- User actions (creating an agent, starting a call) dispatch Laravel controller methods that interact with the ElevenLabs and Twilio SDKs.
- Long-running work (batch calls, data extraction, webhook processing) is pushed onto the Laravel queue and processed by a background queue worker.
- Real-time voice conversations happen directly between the user’s browser and ElevenLabs, with Sniko orchestrating the session via signed tokens.
Queue worker
Sniko relies on a persistent queue worker. In production this is managed by Supervisor (see Installation). In local development,composer run dev starts the queue listener alongside the PHP dev server and Vite.
MCP server
An optional Model Context Protocol (MCP) server ships atmcp-servers/unified/ and runs as a separate Node.js process on port 3001. It exposes Sniko’s agent and conversation data to compatible AI tooling. It is managed with PM2 using the included ecosystem.config.cjs.
Next steps
Quickstart
Register an account, add your ElevenLabs key, and launch your first agent.
Installation
Clone the repo, configure your environment, and self-host Sniko.