Quick start

This guide walks through a full path: install the client → create a fingerprint browser → install a sample automation from the website → test run → personalize with agent chat.

Follow the steps in order. Examples use browser automation (Android / iOS are similar; only the device module differs).

1. Register and download the client

1. Open the website, complete registration, then sign in (email + verification code / password).
2. Go to Download client and get the installer for your OS (one package per OS):
   • Windows: x86 / x64 CPUs only (ARM devices such as some Surface / Snapdragon PCs are not supported)
   • macOS: Apple silicon only (M-series; Intel Mac is not supported)
3. Install and open the AutoAI desktop client, then sign in with the same account.

Notes:

• Registration includes free device quota: 1 browser + 1 Android (see your account page). If you exceed quota, you may be unable to turn on the task master switch or run tasks.
• The desktop client runs on a computer. Installing cases from the website also requires a desktop browser.

2. Create a fingerprint browser

Sample automations usually run in a browser environment—prepare one first.

1. Open Browser in the left sidebar.
2. Use Select browser type and prefer Platform browser (built-in free fingerprint browser). You can also pick BitBrowser or AdsPower (install and start those apps first).
3. Click Create browser (often the sidebar +), fill in name, homepage, proxy (optional), category, then submit.
4. Select the new environment, open its detail page, and click Start. To see the screen, turn on the detail-page Cast switch, or open the sidebar Browser cast hall.

Once created, the environment appears in the list and can be used for cases, debugging, and production runs.

3. Install a sample automation from the website

1. On a computer browser, open the website Case Studies (you must be signed in).
2. Pick an official case close to your scenario and open its detail page.
3. Click Open in client. This launches the desktop app. If it is not installed yet, download the client first, then retry from the case page.
4. When the client is signed in, it downloads the package and opens the import page:
   • Automation cases → Automation → Import
   • Workflow cases → Workflow import
   • Content / tools cases → after import, use My creations / tools
5. On the import page, review details and click Start import (confirm overwrite if a same-name project exists).
6. After success, automation projects appear under Automation → My automations (workflows / creations use their own lists).

If launch fails: make sure the client is installed and signed in, then retry Open in client in a desktop browser. If you are signed out, the client asks you to sign in first.

4. Test run

After import, run once to confirm the environment and script basically work.

4.1 Production-style run (create a task group)

1. Open Automation → My automations.
2. Click the project card body (not the “×× automation expert” button at the bottom) to open the create-task-group dialog.
3. Fill in the task group name and form fields, then click Create.
4. In the Confirm submit dialog, review schedule and task count, then confirm.
5. If prompted Turn on the task master switch?, enable top-right Tasks (tooltip: task execution master switch); otherwise queued tasks will not run.
6. You are taken to Task group management to watch progress and results.

Production task-group runs usually consume almost no tokens (tokens are mainly used for AI development / maintenance / troubleshooting).

4.2 If a run fails

There are two different entries—use the right one:

A. Adapt scripts / environment (Development mode)

1. On the card in My automations, click Browser automation expert (or Android / iOS expert).
2. The session opens in Development mode; pick a started debug device when asked (your fingerprint browser).
3. Describe the issue, e.g. “Step 3 failed after import—please inspect via casting and fix it.”
4. Ask the agent to Trial-run script. After the fix, create a new task group from My automations.

B. Fix a failed task group (Troubleshooting mode)

1. Open Automation → Task group management.
2. Use Investigate and fix on the failed group (not the expert button on the store card).
3. The session opens in Troubleshooting mode for that task group.

Official cases may differ from your local environment (language, account type, site version, region). Adapting with an agent is expected.

5. Personalize with agent chat

Once the sample runs, change the flow in natural language—no hand-written scripts needed.

1. Open the project session from the card’s Browser automation expert (Development mode), or from Agents in the sidebar.
2. Describe changes, for example:
   • “Read the outbound copy from column 2 of the spreadsheet”
   • “Wait 3 more seconds after login before clicking Publish”
   • “Add an optional Notes field when creating a task group”
3. When the agent updates task-group inputs, open the side panel Create preview, click Refresh, and keep chatting until the form looks right.
4. Ask for a Trial-run script, then follow §4.1 to create a task group for batch runs.

For building from scratch (recording → pick an agent → ship an automation), see the next guide.

Next

• Build automation from scratch: recording → agent → form → run
• Devices & automation
• Workflows & content
• Website Case Studies
• Website Download · Register · Sign in