How to Build a 24/7 AI Assistant on Alibaba Cloud with OpenClaw and GLM‑4.7

Introduction

OpenClaw has become popular because it runs a high‑privilege AI agent locally, capable of coding, web browsing, file operations, and shell commands. The author wanted to replace the default Qianwen/Baillian models with the GLM‑4.7 model and expose the agent via a Feishu bot for a 24/7 personal assistant.

What is OpenClaw?

OpenClaw (formerly Clawdbot → Moltbot) is an open‑source AI agent that differs from chat‑only bots. It supports local deployment, multi‑platform coverage (Feishu, WhatsApp, Telegram, Discord, Slack), browser control, system access (file read/write, shell execution), persistent memory, and extensible plugins.

Prerequisites

Resources

Alibaba Cloud ECS (2 CPU, 4 GB RAM or higher)

Feishu Open Platform – create a custom app

GLM Coding Plan – API key from Zhipu AI

Server environment

OS: Ubuntu 20.04+ or CentOS 7+

No need to pre‑install Python/Node.js (the OpenClaw image already includes them)

Deploy OpenClaw on the server

1. Choose OpenClaw image

Buy an ECS instance on Alibaba Cloud and open its detail page.

Click “Reset System” → “Apply Image”.

Select the OpenClaw image and confirm.

The image already contains all required dependencies, so no manual environment setup is needed.

2. Open required ports

# One‑click open required ports
openclaw onboard

3. Configure OpenClaw

Run openclaw onboard and choose “Yes → QuickStart”. The image will finish the installation automatically.

4. Switch to GLM model

During the model selection step, the image defaults to Qianwen/Baillian. Replace it with a GLM version:

Select the GLM model version (e.g., glm‑4‑plus, glm‑4‑7).

Enter the API key obtained from the Zhipu Open Platform.

Skip the Feishu channel because the image already includes the Feishu plugin; selecting it again would cause a conflict.

Create Feishu bot application

5.1 Create a custom app

Visit https://open.feishu.cn and log in.

Click “Create custom app”.

Fill in the app information: name = OpenClaw, description = personal AI assistant, type = Bot.

5.2 Set permissions

Copy the following JSON into the “Batch Import” field of the permission management page:

{
  "scopes": {
    "tenant": [
      "aily:file:read",
      "aily:file:write",
      "application:application.app_message_stats.overview:readonly",
      "application:bot.menu:write",
      "contact:user.employee_id:readonly",
      "corehr:file:download",
      "event:ip_list",
      "im:chat.access_event.bot_p2p_chat:read",
      "im:chat.members:bot_access",
      "im:message",
      "im:message.group_at_msg:readonly",
      "im:message.p2p_msg:readonly",
      "im:message:readonly",
      "im:message:send_as_bot",
      "im:resource"
    ],
    "user": [
      "aily:file:read",
      "aily:file:write",
      "im:chat.access_event.bot_p2p_chat:read"
    ]
  }
}

5.3 Obtain credentials

App ID (e.g., cli_xxxxxxxxxxxxx)

App Secret – a secret string

5.4 Configure Feishu channel in OpenClaw

Enter the App ID and App Secret in the OpenClaw channel configuration page.

Save and restart the service.

Testing

6.1 Check service status

openclaw status

The command should display a running status similar to the screenshot below.

6.2 Feishu bot test

Open Feishu and search for the newly created bot.

Send a message such as “Hello, introduce yourself”.

The bot replies using the GLM model.

Common issues

Q1: Duplicate Feishu plugin warning

Cause: manual installation conflicts with the built‑in plugin or the install command was executed twice.

Solution:

# Stop the service
openclaw gateway stop

# Remove user‑level duplicate plugin
rm -rf /root/.openclaw/extensions/feishu

# Enable built‑in plugin
openclaw plugins enable feishu

# Restart gateway
openclaw gateway restart

Q2: Bot does not reply

Run openclaw status to ensure the service is up.

Confirm the Feishu app has been published.

Check logs with openclaw logs.

Q3: GLM model call fails

Verify the API key is valid and the account has sufficient balance.

Check that the model name (glm‑4, glm‑4‑plus, etc.) is correct.

Inspect the Zhipu platform status page for outages.

Advanced usage

Add persistent memory by storing conversation history in a vector database.

Enable streaming responses for a typewriter‑like effect.

Develop custom plugins to extend AI capabilities.

Deploy multiple channels (Feishu, Telegram, Slack) simultaneously.

Conclusion

You now have a 24/7 AI robot running on Alibaba Cloud, powered by GLM‑4.7, and integrated with Feishu for instant conversation.