Agent Session Quickstart

Connect to your agent and have real-time video conversations using the D-ID Client SDK. The SDK handles WebRTC streaming and provides a simple interface for chat.

Create a client key

Generate a client key for your agents. This key is used by the SDK to authenticate frontend connections.

curl -X POST "https://api.d-id.com/agents/client-key" \
  -H "Authorization: Basic <YOUR KEY>" \
  -H "Content-Type: application/json" \
  -d '{
    "allowed_domains": ["http://localhost:3000", "https://yourdomain.com"]
  }'

{
  "client_key": "Y2xpZW50X2tleV9hYmMxMjM0NTY3ODk...",
}

Save the client_key — you'll use this in your frontend code. The key only works from the allowed origins you specified.

Install the SDK

Add the D-ID Client SDK to your frontend project.

npm install @d-id/client-sdk

Initialize the agent manager

Import the SDK and create an agent manager with your credentials. Connect the HTML video element to the returned video object returned from the SDK

import * as did from "@d-id/client-sdk";

const agentId = "agt_abc123";
const auth = { type: "key", clientKey: "<YOUR CLIENT KEY>" };

// HTML video element to display the agent
const videoElement = document.getElementById("agent-video");
let srcObject;

const callbacks = {
  onSrcObjectReady(value) {
    videoElement.srcObject = value;
    srcObject = value;
  },
  onConnectionStateChange(state) {
    console.log("Connection state:", state);
  },
  onNewMessage(messages, type) {
    console.log("Messages:", messages);
  }
};

const agentManager = await did.createAgentManager(agentId, { auth, callbacks });

Connect and chat

Establish the WebRTC connection and send messages to your agent.

// Connect to the agent
await agentManager.connect();

// Send a chat message - the agent will respond with streaming video
await agentManager.chat("Hello! What can you help me with?");

// When done, disconnect
await agentManager.disconnect();

The video element in the HTML

<video id="agent-video" autoplay playsinline></video>

The agent responds with real-time streaming video. Use agentManager.chat() for conversational messages or agentManager.speak() to make the agent say specific text.