Docs

Volcano Engine Function as a Service

Computer Use Agent

Deploying Computer Use Agent application with one click

This article shows you how to use Function as a Service Application Center to deploy your own Computer Use Agent application with one click.

Background information

Application introduction

Computer Use Agent is an intelligent agent service launched by the Volcano Engine based on ByteDance's Doubao large model, facing end users and AI application developers. The service has built-in self-developed out-of-the-box MCP protocol-driven Computer Use capabilities, supporting mouse control, keyboard control, and screenshot control, and other capabilities for Windows, Linux remote virtual computers. This service easily realizes remotely controlling end points (computers), completing work such as cutting and editing videos, creating PowerPoint slides, and operating and maintaining self-media accounts, etc.

Runtime

Native Python 3.12, Native Node.js 20.x Runtime.

Associated cloud services

warning

The fees for using associated cloud services will be incurred. For billing details, see the billing documentation of the associated cloud services.

Computer Use Agent backend

veFaaS: Provides a large model-driven computer to automate the execution of tasks, real-time viewing of desktops, and interfaces for operating the computer keyboard and mouse.

API Gateway (APIG): exposes APIs.

Elastic Compute Service: Provides Computer instances that execute specific tasks output by the large model.

Virtual Private Cloud: Connects the network between veFaaS and Computer instances.

NAT gateway: Applications access the public network through the NAT gateway.

Public network IP: Applications need to associate the public network IP (EIP) to use the NAT gateway.

Torch Log Service (optional): Provides one-stop services for logs-like data.

Cloud Monitor (optional): Support viewing function basic monitoring via cloud monitoring.

Computer Use Agent frontend

veFaaS: Computer Use Agent interaction console.

API Gateway: Access the console through the API Gateway.

Prerequisites

All the cloud services associated with the application have been activated.

Virtual Private Cloud (VPC), subnets, and security groups have been created in the regions where the applications are pending deployment. For detailed operations, refer to Creating Virtual Private Cloud, Creating subnet, and Creating security group.

IAM users need to ensure that they have authorized VeFaaSFullAccess, APIGFullAccess, STSAssumeRoleAccess, ECSFullAccess, TLSFullAccess, VPCFullAccess policies. If one of the preceding policies is not attached, contact the root user or the administrator for authorization. For specific operations, refer to Creating user and authorize.

The API Gateway instances of the Serverless gateway type that have been created, and ensure that they have the same Virtual Private Cloud (VPC) and subnets as your applications. For detailed operations, refer to Creating instances.

If you use veFaaS for the first time, make sure that the cross-service authorization is complete. If the authorization is not complete, the cross-service authorization page will appear, which provides guidance on authorization.

Procedure

In the top navigation bar, select the target region.

tip

Only the China (Beijing) region supports the application.

In the left navigation pane, click App > App plaza.

On the App plaza page, find the Computer Use Agent template and click Create app.

On the Create application page, specify the parameters of the application.

Basic information

Parameter	Description
Selected template	Displays the selected template. This scenario is a Computer Use Agent.
Name	Specify a custom name for the application based on the naming rule prompted by the system. You can also directly use the application name automatically generated by the system.
Description	Enter the description of the application. This parameter can be left empty.
IAM role	The Application Center provides the system default role ServerlessApplicationRole, which is granted permissions on some cloud resources that the Application Center needs to access. If your user roles do not contain this role, this role will be automatically added. If your user roles contain this role, the system will automatically detect whether all the policies required for the application are attached to this role. If a required policy is not attached, the console will display the name of the policy. You can click Add policy to add the policy. warning Make sure that the IAMReadOnlyAccess policy is attached to IAM users. Otherwise, they will not be authorized to obtain information about the ServerlessApplicationRole role. If IAM users need to further add policies for the ServerlessApplicationRole role, you need to ensure that the IAMFullAccess policy is authorized. For more role-related operations for IAM users, refer to IAM Documentation.
Associated cloud services	Display the names, activation status, and activation purposes of all the cloud services that the current application needs to use. warning For more information about possibly incurred fees and account balance requirements when you use the associated cloud services, see veFaaS billing, APIG billing, VPC billing, ECS billing, TLS billing, Cloud Monitor billing, NAT Gateway billing, and EIP billing.

Resource configuration

Resource configuration modules support the following two configuration methods:

One-click automatic configuration: The system will automatically complete the basic configuration for you, and you only need to fill in the AK/SK information according to the use demand. When the AK/SK is not filled in, the application will create a Linux ECS instance by default, and you need to manage the instance in the ECS console; after filling in the AK/SK, managing the instance in the application is supported.

Custom configuration: You need to customize the configuration for the application parameters in the table below.

Parameter	Description
Basic information
Region	The region where the application and its associated cloud resources are located.
VPC	Select the VPC where you create Computer Use Agent application and cloud-dependent resources.
Subnet	Among the subnets under the selected VPC, select the subnet in which you expect to deploy Computer Use Agent. Only support selecting one subnet.
Security group	Select a security group where you create Computer Use Agent application and cloud-dependent resources.
Large Model Configuration
API Key	Resource API Key for computers to use intelligent agent services. Support selecting from the list of the existing API Key under the current account, or you can also customize the API Key you enter. For obtaining method, refer to API Key Management. The environment variable Key (Env key) corresponding to the Model Resource API Key is: ARK_API_KEY. After creating the application, you can view or modify the model resource API Key through ARK_API_KEY.
GUI Agent model	The doubao-1.5-thinking-vision-pro and doubao-1.5-ui-tars-250328 models are provided by default. The user can choose the model after the Computer Use Agent starts.
Agent Configuration
API Access Key	Fill in the Access Key ID and Secret Access Key (AK/SK) information according to the use demands. When not filled in, the application will create a Linux ECS instance machine by default, and you need to manage the instance in the ECS console; after filling in AS/SK, managing ECS instances in the application is supported.
Access configuration
APIG instance	Select the prepared APIG instance of the serverless type. APIG trigger-related resources will be automatically created for the APIG instance, including services, Upstreams, and routes.
Application Performance Monitoring Configuration	Turn on APMPlus monitoring service according to demands, after which the function monitoring data will be delivered to APMPlus monitoring service, which provides indicators, links, and logs, and other integrated monitoring services, draws the monitoring dashboard, and supports Volcano Ark high code applications and LangChain and other intelligent agent monitoring. For more relevant description of features, refer to APMPlus Documentation.
Memory Configuration	For Computer Use Agent to have long-term memory feature, select to bind the cloud search OpenSearch instance to the Computer Use Agent. In the cloud search console, create a cloud search OpenSearch instance that has the same VPC as Computer Use Agent, and set the password.

Click Confirm to create and deploy the application.

The system will automatically jump to the application details page, and you can view the application deployment status and logs in real time.

Result validation

After the application deployment is successful, under the Basic Information module on the Application Details tab, click Access Entry Point Details to enter your own Computer Use Agent application interaction page.

On the Computer Use Agent application interaction page, select the operating system type you need. Start the sandbox, enter the task to view the remote computer operations, execution steps, and results driven by the large model in real time.

Last updated: 2026.08.19 14:06:06