All Products
Search
Document Center

AgentBay:AgentBay

Last Updated:Aug 24, 2026

AgentBay is a cloud-native execution platform for AI agents. Developers can use it to automate browsers, desktop applications, mobile devices, and code runtimes. By providing isolated, elastically scalable cloud environments, AgentBay solves common problems with local automation scripts, such as inconsistent environments, high local resource consumption, IP restrictions, and security risks.

Why choose AgentBay

  • Automate across diverse scenarios

    • It overcomes the limitations of local agent runtime environments.

    • Provides four core environments for agent development: browsers, code spaces, cloud computers, and mobile devices. These environments cover web automation, code compilation, desktop operations, and mobile app control.

  • Integrate with flexibility

    • Lowers the development barrier and simplifies integration.

    • Offers SDK and MCP integration methods. Experts can perform deep customizations, while others can get started quickly.

  • Run tasks securely in isolation

    • Addresses data security issues and prevents local environment contamination.

    • Uses enterprise-grade secure containers with strict permission isolation, ensuring zero intrusion into your local environment. Supports sandbox process allowlists/blocklists, execution detection, and cross-platform data persistence for state continuity.

  • Scale on cloud infrastructure

    • Reduces the burden of resource management and O&M.

    • Built on Alibaba Cloud infrastructure, AgentBay delivers elastic scaling in seconds and supports thousands of concurrent operations. You can focus on agent innovation instead of infrastructure management and O&M.

Core services

Browser Use

Programmatically control cloud-based browsers to perform tasks such as web scraping, automated web testing, and form filling.

Use cases

  • Data collection: Collect e-commerce product information, industry news, and public data.

  • Automated testing: Simulate user actions to run functional, compatibility, and regression tests on web applications.

  • Process automation: Automatically fill out online forms, enter data, and complete approval workflows.

  • E-commerce operations: Automate product listing, order processing, and price monitoring to improve operational efficiency.

Key features

  • High compatibility: Compatible with the API styles of automation libraries, such as Playwright and Puppeteer, and major agent development frameworks.

  • Multi-environment support: Supports browser operations in mobile, desktop, and browser-only sandbox environments.

  • Flexible extension: Supports all browser running modes, from headless to headed. Also supports loading browser extensions.

Computer Use

Programmatically control a cloud computer environment (Windows or Linux) to operate desktop applications, run automated software tests, and process documents.

Use cases

  • Desktop application testing: Run automated functional and performance tests on Windows and Linux client software.

  • Office automation: Automatically process Excel reports, send and receive emails, and operate internal systems like ERPs.

  • System administration: Perform batch tasks such as software installation, environment configuration, and system maintenance.

Key features

  • Multi-platform support: Supports Windows and Linux operating system images. You can also use a custom image.

  • Secure isolation: Based on VPC network isolation, each task runs in a separate image environment to ensure enterprise data security.

  • Access control: Provides fine-grained access control for cloud computers, including network controls and application allowlists.

Code Space

Provides a cloud-based development environment with pre-installed toolchains. Use it to compile code, run scripts, test APIs, and run CI/CD pipelines.

Use cases

  • Cloud development environment: Use it as a ready-to-use remote development environment without needing to configure a local one.

  • Automated builds and tests: Run code compilation, unit tests, and integration tests in a CI/CD pipeline.

  • Secure code execution: Run untrusted or resource-intensive code in an isolated sandbox.

Key features

  • Pre-installed toolchains: The environment comes with common languages, frameworks, and build and debug tools, ready to use out of the box.

  • Environment consistency: Each run uses a baseline image, ensuring identical development and testing environments.

Mobile Use

Provides a cloud-based Android environment to simulate real phone behavior for tasks such as mobile app testing and app automation.

Use cases

  • Mobile application testing: Run automated functional, compatibility, and performance tests on Android applications.

  • Mobile office: Automate sending and receiving instant messages, summarize message content, and draft polished replies. Automatically create, edit, and convert file formats for mobile documents.

  • Game automation: Perform automated in-game tasks, such as auto-farming or completing quests.

Key features

  • Native mobile support: Supports native device capabilities, such as tapping, swiping, and simulating GPS location and camera input.

  • Multi-device simulation: Supports multiple Android OS versions and screen resolutions.

Billing

AgentBay uses a billing model that combines basic entitlement packages, subscription entitlement packages, and pay-as-you-go. You are billed based on your usage of the following resources and features:

  • Concurrent session licenses: The maximum number of concurrent sessions.

  • Network bandwidth: The inbound network bandwidth consumed during session scheduling.

  • Storage: Used for persisting context data and session replay records.

  • Images: Includes system images and any custom image you create.

  • Points: The number of tokens consumed by the built-in ToolUseAgent when you use AgentBay's AI-related APIs.

  • Advanced features: device fingerprint, and other advanced features.

For more information, see Billing.

FAQ

Multi-user isolation

AgentBay supports user isolation through its session mechanism. You can create a separate session for each end user. Each session runs in an independent sandbox with isolated resources and data.

Built-in model capabilities

Yes. AgentBay includes a built-in ToolUseAgent that orchestrates tools, including those with model capabilities. For this reason, using AgentBay consumes points.