Docs

Volcano Engine Function as a Service

Computer Use Agent

Computer Use Agent Introduction

This topic describes the Computer Use Agent application.

Application introduction

Computer Use Agent is an intelligent agent service launched by Volcano Engine based on ByteDance's Doubao large model, designed for 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 enables you to easily control remote computers to perform tasks such as video editing, PowerPoint creation, and social media account management.

Direct experience

In the top navigation bar, select the target region.

tip

Currently only China (Beijing) region supports App plaza.

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

Find the Computer Use Agent on the Application Square page and click Free Experience to experience it directly.

Process architecture

Associated models and cloud products

Model

Related service	Description	Billing description
Doubao-1.5-thinking-vision-pro-250428	A brand-new visual & deep thinking model, equipped with stronger general multi-modal understanding and reasoning capabilities, and interacts with Sandbox through human-like capabilities including perception, reasoning, and actions, etc.	Model service billing
Doubao-1.5-UI-TARS	A native GUI-oriented Agent model that interacts with Sandbox through human-like capabilities such as perception, reasoning, and actions, etc.
Doubao-embedding	The semantic vectorization model is mainly oriented to the usage scenario of vector retrieval, which is used to retrieve the recorded user context. This model is used only by Computer Use Agents that bind cloud search instances.

Cloud service

Related service	Description	Billing description
ECS	Elastic Compute Service (ECS) is a collection of resources composed of CPU, memory, and cloud disks, each of which is logically related to a computing hardware entity in the data center. You can apply for resources of corresponding size and different specifications according to your own needs for running different business loads	Elastic Compute Service billing
APIG	Volcano Engine API Gateway (APIG) is a cloud-native, highly scalable, and highly available cloud-based gateway hosting service	API gateway billing
veFaaS	veFaaS is a fully managed and Serverless computing platform that is driven by events. Support quickly creating and deploying functions, and integrating middleware and database products on the cloud according to demands.	Function as a Service billing
VPC	Virtual Private Cloud (VPC) builds an isolated, self-configured virtual network environment for resources on the cloud. You can customize Internet Protocol address segments, security groups, and routing policies, and other network features in Virtual Private Cloud to manage resources on the cloud simply and efficiently.	free
NAT gateway	Applications access the public network through the NAT gateway.	NAT gateway billing
EIP	The application using the NAT gateway needs to associate a public network IP.	Public network IP billing
Torch Log Service (optional)	Provides one-stop services for logs-like data.	Torch Log Service billing
Cloud monitoring (optional)	Support viewing function basic monitoring via cloud monitoring.	Cloud monitoring billing
Application performance monitoring full link version (optional)	If you need application monitoring service, you can select to turn on application performance monitoring full link version (APMPlus) product. After turning on this product, the monitoring data of functions will be delivered to the APMPlus monitoring service, which provides integrated monitoring service covering indicators, links, and logs, etc.	Application performance monitoring full link version billing
Cloud search service (optional)	For Computer Use Agent to have long-term memory feature, select to bind the cloud search OpenSearch instance to the Computer Use Agent.	Cloud search service billing

Technology implementation

WebUI: It is responsible for displaying the desktop status of the computer instance in real time, and interactively displaying with the Agent the execution steps and screenshots, and managing the computer instances.

Computer Use Agent backend core components:

Linux/Windows Computer management components: Responsible for creating and deleting computer instances and managing the desktop display of instances.

Task execution components: Responsible for interacting with the large model to obtain execution steps, and operating computer instances according to response results, outputting execution steps in real time, making screenshots in every step, and updating task status.

The Computer Use Agent and Web UI are deployed in veFaaS, and provide task requests and result output via an APIG trigger.

Last updated: 2026.08.19 14:06:06