MAGIC product overview

Updated at:
Copy as MD

MAGIC is an enterprise-grade multi-agent governance and collaboration platform from Alibaba Cloud. It helps enterprises centrally create, orchestrate, and monitor AI Workers (agents), and manage model configurations and team permissions. This document introduces the product positioning, core features, use cases, and supported regions of MAGIC.

What is MAGIC?

MAGIC is a multi-agent governance and collaboration platform for enterprises, offering a unified console to create and manage AI Workers. With MAGIC, you can connect to multiple AI model providers, integrate external tools (MCP services), and monitor the runtime status of AI Workers in real time.

MAGIC supports multi-member, multi-role team collaboration and permission management to meet enterprise security and compliance requirements.

Core features

Feature

Description

Instance management

Create and manage MAGIC instances, which act as resource isolation boundaries. This allows you to allocate resources independently for projects or business lines.

Worker management

Create and manage AI Workers. This feature supports multiple operating modes for various business scenarios.

Model management

Connect to multiple AI model providers and centrally manage model configurations. You can switch models between AI Workers as needed.

Team collaboration

Create teams, manage member permissions, and support multi-role collaboration for enterprise-grade access control.

MCP service

Integrate and manage external tools based on the Model Context Protocol (MCP), providing AI Workers with extended tool-calling capabilities.

Monitoring dashboard

View real-time AI Worker usage, model call volumes, and token consumption trends. Supports usage analysis and anomaly detection.

Worker template

Create reusable worker configuration templates to quickly deploy similar types of agents in bulk, reducing repetitive configuration.

Use cases

  • Enterprise digital employees: Employees initiate tasks through enterprise IM applications (such as DingTalk, Feishu, and WeCom). MAGIC then dispatches an agent team by department. The agents use MCP services to access enterprise systems like CRM, OA, and ERP. The entire process supports IDP authentication, audit logs, and cost attribution.

  • Agent team as a service: Platform administrators create pools of agent teams based on roles (such as R&D, operations, data, and security). Business teams can then request on-demand access to the appropriate teams through role-based access control (RBAC). This enables permission isolation, independent quota-based billing, and unified platform maintenance.

  • SaaS agent team enablement: SaaS vendors route each tenant to a dedicated agent team. Using a permission isolation and access control engine, vendors can assign combinations of callable MCP services based on customer authorization policies. This enables multi-tenant data isolation and independent usage metering.

  • Management of existing agents: Using the MAGIC unified governance platform, enterprises can incorporate heterogeneous agents (such as OpenClaw, QwenPaw, Claude Code, and in-house agents) into a unified team. This allows you to manage these agents without modification, consolidating and enabling cross-team reuse of agent assets, such as MCP services, prompts, and knowledge bases.

Supported regions

MAGIC is currently available in the following regions:

Region name

Region ID

China (Hangzhou)

cn-hangzhou

China (Beijing)

cn-beijing

Asia Pacific SE 1 (Singapore)

ap-southeast-1