Build your AI-powered
customer agent

Overview

Everything your business needs to deploy a live AI agent on your website.

Installation

Get up and running in under 5 minutes.

General

Basic appearance and language settings for your widget.

My Agent

Configure your AI agent's personality, voice, and knowledge.

System Prompt Tips

Your system prompt defines how the agent behaves. Be specific about what it should and should not do. Include your business name, tone of voice, topics to focus on, and any rules for taking actions.

Knowledge Base Tips

Include everything your agent needs to answer visitor questions accurately: services, pricing, hours, location, contact details, and FAQs. The more complete your knowledge base, the better your agent performs.

Voice Reference

30 voices available — use the TTS Preview button in the My Agent tab to audition any voice before going live.

Automation

Connect your agent to n8n, Make, Zapier, or any webhook to take real actions during conversations.

How It Works

When a visitor asks the agent to do something (book an appointment, check availability, send a quote), the agent collects the required information and calls your webhook URL. Your automation platform handles the action and returns a result. The agent relays the result back to the visitor.

Webhook URL

Paste your webhook URL from n8n, Make, or Zapier into the Webhook URL field and click Save Webhook.

Agent Tools

Tools define what actions your agent can perform. Each tool has a name, a description that tells the agent when to use it, a list of parameters to collect from the visitor, and a thinking message shown while the webhook is processing.

Widget

Customize the appearance and behavior of the floating widget on your site.

Dashboard

Monitor your usage and review recent conversations.

Voice Call Flow

What happens from the moment a visitor starts a call.

Audio Pipeline

All voice communication runs over a plain WebSocket — no WebRTC. The browser captures and sends 16 kHz PCM audio (base64-encoded) to the backend, which processes it through StudioMeta's real-time AI engine. The agent's audio response (24 kHz PCM, base64-encoded) is streamed back to the browser in real time.

Chat Flow

How the chat widget processes messages and triggers actions.

The chat widget maintains full conversation history within the session. If an action is required, the agent collects all necessary information before calling the webhook — it will not call the tool until it has everything it needs.

Automatic Lead Capture

In both voice and chat modes, the agent automatically collects visitor name, phone, and email mid-conversation. Captured leads appear instantly in your Leads tab — no extra configuration required.

Webhook API Reference

The data format sent to your automation platform and what it expects back.

Tool Call Request

Sent to your webhook when the agent calls a tool — during both voice and chat.

{
  "event": "voice_tool_call",        // or "chat_tool_call"
  "tool_name": "your_tool_name",
  "tool_args": {
    // parameters your tool defined
    // e.g. name, email, phone, date, service...
  },
  "call_id": "call_abc123",
  "license_key": "SM-XXXX-XXXX-XXXX",
  "business_name": "Your Business Name",
  "site_url": "https://yourwebsite.com",
  "timestamp": "2026-05-13T13:00:00.000Z"
}

Call Completed Request

Sent to your webhook automatically when a voice call ends, for logging or email summaries.

{
  "event": "call_completed",
  "license_key": "SM-XXXX-XXXX-XXXX",
  "customer_email": "[email protected]",
  "duration_min": 3,
  "transcript": "user: ...\nassistant: ...",
  "ended_at": "2026-05-13T13:05:00.000Z"
}

Required Response Format

Your webhook must return a JSON object with a result field. The agent reads this and relays it to the visitor.

{
  "result": "Your appointment has been booked for Saturday at 3:00 PM."
}

Plans & Limits

Voice minutes and chat messages reset on the first of each month.

Troubleshooting

Common issues and how to resolve them.