Quickstart

Create a conversational video agent with the REST API, then drop it into any web page. This uses platform defaults for every advanced setting: you’ll tune those later.

Prerequisites

A TruGen API key from the Developer Portal
An avatar ID: the Avatar Gallery has stock avatars you can use right away

1. Create an agent

Send a POST /v1/ext/agent with a name, prompt, and avatar. Everything else falls back to platform defaults (Groq LLM, ElevenLabs voice + STT, English, no tools).

curl --request POST \
  --url https://api.trugen.ai/v1/ext/agent \
  --header 'Content-Type: application/json' \
  --header 'x-api-key: YOUR_API_KEY' \
  --data '{
    "agent_name": "My First Agent",
    "agent_system_prompt": "You are a friendly assistant that answers questions in one or two short sentences.",
    "avatars": [
      { "avatar_key_id": "665a1170" }
    ]
  }'

The response returns an id: save it for step 2.

Response

{
  "id": "agt_12345example",
  "message": "Agent created successfully"
}

Need every configurable field? See the full configuration reference.

2. Embed the agent

Pick the integration that matches your stack.

React
iFrame
JavaScript SDK

Install the widget:

npm install @aiteammate/agent-widget

Render it with your agent ID:

import "@aiteammate/agent-widget/styles.css";
import { TrugenAgentWidget } from "@aiteammate/agent-widget";

export default function AgentPage() {
  return (
    <TrugenAgentWidget
      agentId="<AGENT_ID>"
      userName="Jane"
      userId="user_123"
      context="Jane opened the pricing page"
    />
  );
}

Drop this into any HTML page:

<iframe
  src="https://app.trugen.ai/embed/<AGENT_ID>?username=Jane&id=user_123&context=Jane%20opened%20pricing"
  width="100%"
  height="600"
  frameborder="0"
  allow="camera; microphone; autoplay; display-capture"
></iframe>

For custom UIs and full session control:

npm install @trugen/js-sdk

import { createClient, TruGenEvent } from "@trugen/js-sdk";

// Get a session token from your backend (never expose your API key to the browser)
const token = await fetch("/api/session-token")
  .then((r) => r.json())
  .then((d) => d.token);

const session = await createClient({ token });
session.on(TruGenEvent.VIDEO_STREAM_STARTED, (track) => {
  track.attach(document.getElementById("avatarVideo"));
});
await session.connect();

See the JavaScript SDK quickstart for a working end-to-end example.

Open the page in a browser and grant camera + mic permissions. Your agent joins the call and starts speaking.

What just happened

Step 1: Agent created. Behind that single API call, TruGen provisioned a persistent agent with your prompt and avatar, wired up the default pipeline (Groq LLM, ElevenLabs STT + TTS, English, turn detection on), and returned an ID. That ID is your handle to the agent everywhere: widget, iframe, SDK, external meetings, and API. Change the agent’s behavior anytime with PATCH /v1/ext/agent/{id} and the next session picks it up. No redeploy of your app required. Step 2: Live conversation. The widget or iframe traded your API key for a short-lived session token, opened a WebRTC connection to TruGen’s edge, streamed the user’s microphone in, and streamed the rendered avatar video back. Every turn ran through the full pipeline (capture → STT → Hawkeye vision → LLM → TTS → Huma render) server-side in under a second. You wrote zero WebRTC code, managed zero media state, and never exposed your API key to the browser.

Next steps

Agent concepts

What an agent is, how the conversation loop works, and everything you can configure.

Prompting

Write system prompts that behave well in a spoken, real-time setting.

Introduction

Models

AI Teammates (Avatars)

Text to Video

Prerequisites

1. Create an agent

2. Embed the agent

What just happened

Next steps

Agent concepts

Prompting

​Prerequisites

​1. Create an agent

​2. Embed the agent

​What just happened

​Next steps

Agent concepts

Prompting

Prerequisites

1. Create an agent

2. Embed the agent

What just happened

Next steps