Genesys Cloud and Amazon Connect CCP monitoring for agent desktops

What uptime checks miss

A ping tells you the sign-in page answers. It can't tell you whether an agent gets from that page to a working desktop. These failures leave agents waiting while the URL still returns HTTP 200:

Agent desktop monitoring catches these because it does what an agent does at the start of a shift. It opens the desktop, signs in and waits for the agent UI.

How it works

1Create a test agent and an environment

Set up an agent account just for monitoring. In VerLake, create an environment with the desktop URL and the agent's credentials. They reach the script as process.env. Add one environment per region or business unit, and the same script runs against each.

2Write the sign-in journey

Use the example below, or describe the flow in the editor's AI tab and it drafts the script. The AI assistant uses your organization's own AI API key.

3Test it in the editor

Click Test Run. You get each step, the console output, screenshots and a video replay, so you can see the desktop load before you schedule it.

4Schedule it and set a threshold

Run it every 5 minutes, all day. Set a failure threshold of 2 or 3 so one slow load doesn't page anyone. With a 5-minute interval and a threshold of 2, you hear about a broken sign-in within about 10 minutes.

Example: agent sign-in check

This journey opens the agent desktop, signs in as the test agent and checks that the agent UI appears. It does nothing else. The labels and the final check are placeholders. Open your desktop, look at the sign-in form and the loaded UI, and adjust them to match.

import { test, expect } from '@playwright/test';

// Placeholder selectors: adjust the labels and the final check to match
// your Genesys Cloud or Amazon Connect sign-in page and agent UI.
test('agent can sign in to the desktop', async ({ page }) => {
  await test.step('Open the agent desktop', async () => {
    // e.g. https://login.mypurecloud.com
    // or   https://your-instance.my.connect.aws/ccp-v2
    await page.goto(process.env.AGENT_DESKTOP_URL);
  });

  await test.step('Sign in as the test agent', async () => {
    await page.getByLabel('Username').fill(process.env.AGENT_USER);
    await page.getByLabel('Password').fill(process.env.AGENT_PASS);
    await page.getByRole('button', { name: 'Sign in' }).click();
  });

  await test.step('Agent UI loads', async () => {
    await expect(page.getByRole('main')).toBeVisible({ timeout: 20_000 });
  });
});

If agents sign in through your own identity provider, add those steps between opening the desktop and the agent UI. The SSO example shows how. Once the sign-in check is stable, you can add steps that check the softphone panel and queues render.

What you see when a check fails

VerLake waits for the number of failures in a row you set. Then it alerts you, and it tells you again when the desktop check recovers.

Every run records:

• the step that failed: opening the desktop, signing in, or the agent UI
• how long every step took
• the browser console output and the error
• an AI summary of the likely root cause, using your own AI API key

Runs stay in Run History, so you can compare a failure with the last good run. With the MCP connector and its 16 tools, your AI assistant can tell you whether the check failed overnight.

Pricing

You pay for the seconds a check runs. Queue time is free.

A 20-second sign-in check every 5 minutes uses 172,800 seconds a month, so it fits in Starter. Beyond your plan it's $0.10 per 1,000 seconds, with a spend cap you control. See full pricing.

FAQ