Configure WeChat Service Account Customer Support

Quick configuration enables CloudBase AI to integrate with Service Account customer service to achieve intelligent customer service features.

Prerequisites

• You have registered WeChat official accounts (Service Account)
• You have enabled the CloudBase environment (click to activate)
• You have created an Agent (click to create)

Step 1: Obtain the Service Account AppId

Go to the Service Account backend and view the AppId in Settings & Development > Development Interface Management.

On first entry, you need to become a developer first:

After becoming a developer, you can view the AppId:

Step 2: Authorize Service Account

1. Go to the CloudBase AI console and select the Agent module.
2. Locate the "WeChat Platform" in the "Integration Guide"
3. Select "Service Account" and click "Authorize Now"
4. Enter the AppId obtained in the previous step and click Next.
5. Use the Service Account administrator's WeChat to scan the QR code and complete authorization.

📝 Note: This authorization allows CloudBase to act as an agent for Service Account customer service, enabling AI auto-reply to messages.

Step 3: Development

After authorization is completed, the Agent needs to be modified to support the WeChat platform message format. Refer to the corresponding development documentation based on your Agent type:

• Agents with the agent- prefix: See WeChat Platform Adapter Development Documentation
• Agents with the ibot- prefix: See Integration of ibot- prefixed Agents with WeChat callback messages

Step 4: Testing

After completing development and deployment, you can test the intelligent customer service feature:

1. Open your Service Account conversation window
2. Send any message
3. Wait for the AI Agent to automatically reply

Optimization Suggestions

Improve Response Speed

Add the following settings to the Agent prompt to significantly improve response speed:

• The Agent responds to questions in a minimalist style
• Simplify complex questions and extract the key information
• Strictly limit the length and relevance of responses to avoid redundancy
• Do not output in Markdown format; output plain text directly

⚠️ Note: After changing the prompt, historical conversation records may affect the model's output effectiveness.

Model Selection

• Recommended: Hunyuan model, DeepSeek V3 (fast response speed)
• Not recommended: DeepSeek R1 and other deep thinking models (response time is slow)

Better Experience

For a richer conversation experience (such as Markdown rendering and multi-turn conversation display), you can consider embedding H5 pages or Mini Programs in the Service Account menu.