SDKs and client libraries

Tried ships one client library, @tried/web-sdk, for running web calls in the browser. Server code talks to the REST API over plain HTTPS, so any HTTP client works. The website chat widget is a single script tag.

Browser SDK#

@tried/web-sdk wraps livekit-client. It joins the call's LiveKit room with the token from POST /v3/create-web-call, publishes the microphone, plays the agent's audio and emits transcript and talking events. Its API matches Retell's web client, so front ends written for RetellWebClient switch by changing the import.

Start a web call#

Never put your API key in the browser. Create the web call on your server and send the browser only the access_token and livekit_url.

// Your backend: create the web call and return the token to the browser.
const res = await fetch("https://api.tried.com.au/v3/create-web-call", {
  method: "POST",
  headers: {
    Authorization: `Bearer ${process.env.TRIED_API_KEY}`,
    "Content-Type": "application/json",
  },
  body: JSON.stringify({
    agent_id: "agent_3f1c9a7b2d4e6f8091a2b3c4",
    retell_llm_dynamic_variables: { first_name: "Sam" },
  }),
});
const { call_id, access_token, livekit_url } = await res.json();

import { TriedWebClient } from "@tried/web-sdk";

const client = new TriedWebClient();

client.on("call_started", () => console.log("connected"));
client.on("agent_start_talking", () => setTalking(true));
client.on("agent_stop_talking", () => setTalking(false));
client.on("update", ({ transcript }) => render(transcript));
client.on("call_ended", () => console.log("call ended"));
client.on("error", (err) => console.error(err));

// call from a click handler so the browser allows audio
await client.startCall({ accessToken: access_token, livekitUrl: livekit_url });

// later
client.mute();
client.unmute();
client.stopCall();

A browser can also start web calls without your server by using a public key from Chat Agents, Website widget. Public keys are limited to allowed domains and can require reCAPTCHA. See Deploy.

startCall(config)#

The microphone is opened with echo cancellation, noise suppression and auto gain control.

Methods#

The isAgentTalking property is true while the agent speaks.

Events#

Client options#

new TriedWebClient(options) accepts:

Migrating from Retell's web client#

// before
// import { RetellWebClient } from "retell-client-js-sdk";
import { RetellWebClient } from "@tried/web-sdk";

const client = new RetellWebClient();
await client.startCall({ accessToken });

The default export is also the client class.

Chat widget script#

Add the chat widget to a site with one script tag served by the API. It needs a public key created under Chat Agents, Website widget.

<script src="https://api.tried.com.au/widget.js"
  data-public-key="public_key_..."
  data-title="Chat with us"
  data-color="#4f46e5"></script>

Other attributes: data-api (API base URL) and data-recaptcha-site-key (reCAPTCHA v3 site key). See Chat agents.

Server code#

Tried does not publish a server SDK. Every endpoint is plain JSON over HTTPS with a bearer key, so fetch, requests or any HTTP client works. Every API reference page has cURL, Node and Python examples.

curl https://api.tried.com.au/v2/get-call/call_8d2f0c1e9b7a6543210fedcb \
  -H "Authorization: Bearer $TRIED_API_KEY"

const call = await fetch("https://api.tried.com.au/v2/get-call/call_8d2f0c1e9b7a6543210fedcb", {
  headers: { Authorization: `Bearer ${process.env.TRIED_API_KEY}` },
}).then((r) => r.json());

import os, requests

call = requests.get(
    "https://api.tried.com.au/v2/get-call/call_8d2f0c1e9b7a6543210fedcb",
    headers={"Authorization": f"Bearer {os.environ['TRIED_API_KEY']}"},
).json()

If your existing code uses a Retell server SDK, set that SDK's base URL option to https://api.tried.com.au and pass your Tried API key. The paths it calls are the ones Tried serves; check Migrate from Retell for field and path differences.