Deploy OpenClaw agent PAI-EAS quick start - Platform For AI

OpenClaw is an open-source, self-hostable AI agent framework that transforms a large language model from a simple chatbot into a digital assistant that executes tasks. With natural language commands, it can handle file processing, code writing, and process automation, acting as your digital employee. This topic shows you how to deploy a dedicated OpenClaw agent in one click using PAI-EAS. In addition to calling services from the Alibaba Cloud Model Studio platform, you can directly call your custom model services on PAI-EAS, allowing you to easily create an integrated solution with a dedicated large language model and a dedicated AI agent.

Deploy an EAS service

Log on to the PAI console. Select a region on the top of the page. Then, select the desired workspace and click Elastic Algorithm Service (EAS).
On the Inference Service tab, click Deploy Service. In the Scenario-based Model Deployment section, click OpenClaw Deployment.

Configure the following key parameters:

Parameter	Description
Service Name	Enter a custom service name. Example: openclaw_demo
Image Configuration	Select the latest version.
Resource Configuration	Select a CPU instance type. By default, `ecs.c7a.large` is selected.
Model Settings	Select an OSS path to persist the OpenClaw configuration, such as `oss://examplebucket/openclaw/`.
VPC	You must create and configure the required VPC, VSwitch, and security group. Note OpenClaw requires public network access for features such as web search and browser applications. For optimal performance, select a VPC that is already configured for public network access.

Click Deploy and wait for the service status to change to Running. The deployment is successful when the status is Running.

Configure public network access

An EAS service accesses the public internet through an Internet NAT Gateway. For more information, see Internet NAT Gateway.

1. Create a NAT gateway and bind an EIP

Go to the NAT Gateway - Internet NAT Gateway purchase page.

Billing Method: Select pay-as-you-go.
Region: Select the region where your EAS service is located.
Network and Zone: Select the VPC configured for your EAS service. This setting cannot be changed after creation.
EIP: Select an EIP that is not associated with an instance, or purchase a new one.

2. Configure an SNAT entry

Go to the Internet NAT Gateway page. Find the Internet NAT Gateway that you created. In the Actions column, click Configure SNAT , and then click Create SNAT Entry.

SNAT Entry: Select VPC granularity.
Select EIP: Select the configured EIP.

Launch the WebUI

Click the button in the Invoke/Log/Monitoring column to open the web UI page.
Configure the model. The service defaults to using an Alibaba Cloud Model Studio model in the China (Beijing) region. Obtain an API key for the same region, enter it in the WebUI, and save the configuration. You can also configure an Alibaba Cloud Model Studio Coding Plan or an EAS Model Service. For more information, see Configure multiple model providers.

After you click Save, the key is automatically redacted for security and displayed as __OPENCLAW_REDACTED__. In the left-side navigation pane, click Configuration, select the Settings tab, and switch to Raw mode. In the JSON5 configuration, locate the apiKey field under providers > dashscope and enter your API key.
On the WebUI page, you can start a conversation with your private OpenClaw assistant and assign tasks to it.

On the Chat page of the OPENCLAW GATEWAY DASHBOARD, the left side features a navigation bar with categories such as Chat, Controls, Agents, and Settings, while the right side is the chat session area. At the top, you can use the session ID input box to switch sessions. The bottom provides a message input box and the Send and New session buttons. The assistant supports tool calling. For example, it can use the Exec tool to query weather information and return a formatted Chinese summary.

Configure channels

DingTalk integration

Create a DingTalk bot and obtain its Client ID and Client Secret. For more information, see Create an OpenClaw bot with one click.
In the OpenClaw WebUI, go to Configuration > RAW. In the channels configuration block, find dingtalk-connector, enter the obtained Client ID and Client Secret, and save the configuration.

In the dingtalk-connector configuration block, in addition to entering the clientId and clientSecret, ensure that enabled is set to true. You can also adjust fields such as separateSessionByConversation, groupSessionScope, and sharedMemoryAcrossConversations as needed.

Lark

Go to the Lark Open Platform, click Create Enterprise Self-Built App, enter an application name and description, select an application icon, and then click Create.
In the left-side navigation pane, click Credentials and Basic Information. Copy the App ID (for example, cli_xxx) and App Secret.

In the left-side navigation pane, click Permission Management. Click Batch Import/Export Permissions, paste the following JSON configuration, click Next, Confirm New Permissions, and then click Request Activation.

JSON configuration

{
  "scopes": {
    "tenant": [
      "aily:file:read",
      "aily:file:write",
      "application:application.app_message_stats.overview:readonly",
      "application:application:self_manage",
      "application:bot.menu:write",
      "cardkit:card:write",
      "contact:user.employee_id:readonly",
      "corehr:file:download",
      "docs:document.content:read",
      "event:ip_list",
      "im:chat",
      "im:chat.access_event.bot_p2p_chat:read",
      "im:chat.members:bot_access",
      "im:message",
      "im:message.group_at_msg:readonly",
      "im:message.group_msg",
      "im:message.p2p_msg:readonly",
      "im:message:readonly",
      "im:message:send_as_bot",
      "im:resource",
      "sheets:spreadsheet",
      "wiki:wiki:readonly"
    ],
    "user": ["aily:file:read", "aily:file:write", "im:chat.access_event.bot_p2p_chat:read"]
  }
}

In the left-side navigation pane, click Add App Capabilities. On the Add by Capability tab, find the Bot card and click Configure.
Configure event subscription.
1. In the OpenClaw WebUI, go to Configuration > RAW. In the JSON content, find the feishu channel, add the App ID and App Secret that you obtained, and then click Save. Note: For security, the secret is automatically redacted and displayed as __OPENCLAW_REDACTED__ after you click Save.
2. On the Events and Callbacks page of the Lark Open Platform, confirm that the subscription method is set to Persistent Connection.
3. On the event configuration page, click Add Event, search for the Receive Message event, and then click Confirm Add.
On the Version Management and Publishing page, create a version. Enter an Application Version Number and an Update Description, click Save, and then submit the application for review and publishing.

In the version details, set both the Default Mobile Capability and Default Desktop Capability to Bot.
Add the bot to your favorites. In the Workbench, click Add to Favorites. Search for the bot by its name and click Add. The bot appears in your favorites. You can double-click it to start a conversation.

WeCom

Create a WeCom bot and obtain its Bot ID and Secret. For more information, see How to integrate OpenClaw with WeCom.

In the OpenClaw WebUI, go to Configuration > RAW. In the channels configuration block, add the configuration for WeCom. Replace the botId and secret placeholders with your bot's credentials.

    "wecom": {
      "enabled": true,
      "botId": "${WECOM_BOT_ID}",		// Replace with your actual content.
      "secret": "${WECOM_SECRET}",	// Replace with your actual content.
      "allowFrom": [
        "*"
      ],
      "dmPolicy": "open"
    }

More WebUI configurations

Configure multiple model providers

In the WebUI, select Configuration > RAW. In the JSON, find the models configuration block. Under providers, you can configure multiple model providers:

{ 
  // Other configurations omitted ...
  models: {
    mode: 'merge',
    providers: {
      dashscope: {},
      codingplan:{},
      eas:{},
    },
   },
}

Coding Plan

baseUrl: https://coding.dashscope.aliyuncs.com/v1
apiKey: Go to the or the Coding Plan page to subscribe to the service and obtain your API key for Coding Plan.

providers: {
  codingplan : {
    baseUrl: 'https://coding.dashscope.aliyuncs.com/v1',
    apiKey: '__OPENCLAW_REDACTED__',        // Replace with your actual content.
    api: 'openai-completions',
    models: [
      {
        id: 'qwen3.5-plus',
        name: 'qwen3.5-plus',
        reasoning: false,
        input: [
          'text',
          'image',
        ],
        contextWindow: 1000000,
        maxTokens: 65536,
      },
      {
        id: 'qwen3-coder-next',
        name: 'qwen3-coder-next',
        reasoning: false,
        input: [
          'text',
        ],
        contextWindow: 262144,
        maxTokens: 65536,
      },
    ],
  },
},

EAS Model Service

Obtain the endpoint and token of the EAS service.

baseUrl: The invocation URL of the EAS service. Append /v1 to the URL.
apiKey: The token for your EAS service.

    providers: {
      eas: {
        baseUrl: 'http://xxxxxx.cn-hangzhou.pai-eas.aliyuncs.com/api/predict/quickstart_deploy_xxxx/v1/',
        apiKey: '__OPENCLAW_REDACTED__',
        api: 'openai-completions',
        models: [
          {
            id: 'Qwen3.5-27B',
            name: 'Qwen3.5-27B',
            api: 'openai-completions',
            reasoning: false,
            input: [
              'text',
            ],
            cost: {
              input: 0,
              output: 0,
              cacheRead: 0,
              cacheWrite: 0,
            },
            contextWindow: 1000000,
            maxTokens: 65536,
          },
        ],
      },
    },

DashScope

baseUrl: The default value is https://dashscope.aliyuncs.com/compatible-mode/v1.
apiKey: Enter the API key for the same region as the baseUrl.

  models: {
    mode: 'merge',
    providers: {
      dashscope: {
        baseUrl: 'https://dashscope.aliyuncs.com/compatible-mode/v1',
        apiKey: '__OPENCLAW_REDACTED__',    // Replace with your actual content.
        api: 'openai-completions',
        models: [
          {
            id: 'qwen3.5-plus',
            name: 'qwen3.5-plus',
            reasoning: false,
            input: [
              'text',
              'image',
            ],
            contextWindow: 1000000,
            maxTokens: 65536,
          },
        ],
      },
    },
  },

Configure the default model

In the WebUI, select Configuration > RAW. In the JSON, find the agents configuration block.

agents.defaults.model.primary: Sets the default model.
Model name: The format is provider/model, where provider is the name of the model provider and model is the model ID.

  agents: {
    defaults: {
      model: {
        primary: 'dashscope/qwen3.5-plus',
      },
      models: {
        'dashscope/qwen3.5-plus': {},
        'coding/qwen3-coder-next': {},
        'eas/Qwen3.5-27B': {}
      },
    },
  },