AutoGLM Mobile AI Assistant for PC

A mobile automation tool based on Zhipu AI that controls Android phones to perform various tasks through natural language commands

Bug report

Loading tool, please wait...

If the online tool fails to load or run, try the desktop tool.https://tools.yikeaigc.com/

Tool Usage

Return to old version

Instructions

Software Usage Instructions

Install the ADB environment: Make sure the ADB debugging tool is installed on your computer. The software requires ADB support to run. Please install and configure the ADB environment before first use.
Enable Developer Mode on your phone:
- Go to your phone's "Settings" → "About phone" and tap "Build number" 7 times in a row to enable Developer Mode
- Return to Settings, go to "Developer options", and enable "USB debugging"
Connect your phone device:
- USB connection: Connect your phone to the computer with a data cable, and the software will automatically detect the device
- Wireless connection: Enter the phone's IP address and port number (default 5555), then click the "Connect" button
Configure the API service:
- Select an API provider (Zhipu BigModel or ModelScope)
- Enter the API Key you applied for
- You can click "API Application Tutorial" to see how to obtain one
Enter task instructions: Describe the operation you want to perform in natural language in the task input box, for example: "Open WeChat and send the message hello to File Transfer Assistant".
Execute the task: Click the "Start Execution" button, and the software will automatically analyze the instructions and operate the phone to complete the task. The execution process will be displayed in real time in the log area.

FAQ

A: The software supports more than 50 mainstream Chinese apps, including social and messaging apps such as WeChat and QQ, food delivery apps such as Meituan and Ele.me, e-commerce apps such as JD.com and Taobao, short video apps such as Douyin and Kuaishou, as well as travel apps such as Amap and DiDi. It covers multiple scenarios including social networking, shopping, food delivery, entertainment, and travel.

A: Zhipu BigModel is the official Zhipu AI API. It uses the autoglm-phone model and requires you to apply for an API Key at open.bigmodel.cn; ModelScope is Alibaba Cloud's ModelScope platform. It uses the open-source AutoGLM-Phone-9B model and requires you to apply for an API Key at modelscope.cn. The two offer similar functionality, so you can choose according to your needs.

A: A USB connection is more stable and reliable, and is recommended for first-time use; a wireless connection is more convenient and flexible, and is suitable for experienced users. Wireless connection requires the phone and computer to be on the same local network, and wireless debugging must first be enabled via a USB connection.

A: The software simulates real user actions on the phone, including tapping, swiping, and entering text. When a task is running, the phone screen will show the corresponding operation process, and you can observe the task execution in real time. To stop midway, click the "Stop Execution" button.

A: Possible reasons include: 1) USB debugging is not enabled on the phone; please enable it in Developer options; 2) the data cable only supports charging and does not support data transfer; please replace the data cable; 3) you did not tap Allow on the authorization prompt that appeared when connecting the phone; 4) the ADB environment is not correctly installed. Please check the above issues one by one, then click "Refresh Devices" and try again.

A: Root access is not required. The software communicates with the phone through the ADB debugging interface; you only need to enable USB debugging to use it. The software will automatically install the ADB Keyboard input method to support Chinese input. Please allow the installation on the phone the first time it runs.

A: Task execution may fail for several reasons: 1) Unstable network connection causing API calls to fail; 2) The task instructions are not clear enough; 3) The phone interface does not match expectations. Recommendation: Click "Stop Execution" to terminate the current task, check the log information to identify the issue, optimize the task instruction description, and then run it again. The clearer and more specific the task instructions are, the higher the success rate.