Alibaba’s DingTalk launches Wukong, an AI-native work platform for enterprise agents
On March 17, Alibaba’s DingTalk announced Wukong at its AI DingTalk 2.0 annual launch event, positioning it as the “world’s first enterprise AI-native work platform.” The company opened invite-only testing for a standalone Wukong app and said the platform will also be embedded inside DingTalk for both PC and mobile use. DingTalk claims the move turns enterprise AI from tool-level add-ons into a 24/7 execution layer by letting agents operate directly within corporate accounts, permissions, and workflows.
Wukong launch: standalone app plus native DingTalk integration
DingTalk’s rollout plan is twofold: Wukong ships as an independent application that enterprises can test immediately, while a built-in version is slated to run inside DingTalk itself. The dual path matters because it lets companies try the product quickly without waiting for a full in-app deployment, while preserving the long-term goal of deep integration. DingTalk also emphasized that Wukong will work across PC and mobile clients, signaling a platform that is meant to follow employees across devices rather than sit in a single desktop workflow.
How it works: “CLI-ified” DingTalk and agent skills
A key technical claim from media coverage is that DingTalk’s core has been “CLI-ified,” meaning agents can call DingTalk’s functions natively rather than relying on brittle UI automation. The company says this gives AI agents access to thousands of enterprise capabilities—such as messaging, scheduling, approvals, and system integrations—through a command layer that works across PC and mobile. DingTalk also emphasizes 24-hour task execution, signaling a push toward continuous, agent-driven workflows rather than single-step copilots. In addition, Alibaba’s ToB ecosystem is expected to plug in as “skills,” creating a marketplace of enterprise functions that agents can invoke as part of longer, multi-step processes.
From copilots to workflow automation
Wukong’s positioning points to a shift from question-and-answer bots to workflows that can actually execute business tasks. The trend echoes enterprise agent momentum in China, including AI Agents Drive Enterprise Data Infrastructure Revolution While Chinese Entrepreneurs Cash In on OpenClaw Boom. Instead of deploying multiple point tools, organizations can run agents within a single platform that understands corporate roles, data access, and workflow triggers. That matters in China’s enterprise market, where compliance, security, and auditability are central to adoption. By anchoring agents inside DingTalk’s permission system, Alibaba is trying to make AI execution feel like a controlled extension of existing enterprise operations rather than a standalone experiment.
Why it matters for DingTalk’s enterprise ecosystem
DingTalk already sits at the center of many Chinese companies’ collaboration workflows, making it a logical front door for enterprise AI. By introducing Wukong as both a standalone app and an embedded layer, Alibaba is effectively turning DingTalk into an operating system for agent-driven work. The promise is lower integration overhead: instead of stitching together multiple AI tools across business systems, companies can route execution through a single, permission-aware platform. If that model works, it strengthens DingTalk’s ecosystem lock-in and encourages partners to build more enterprise-grade skills on top of the platform. The launch also aligns with Alibaba’s move to centralize its foundation-model strategy after Qwen’s exit, signaling tighter integration across the company’s AI stack.
Market context in China
Alibaba’s launch lands as China’s AI market scales up. According to the China Academy of Information and Communications Technology (CAICT), the country’s core AI industry exceeded RMB 900 billion in 2024 and is projected to surpass RMB 1.2 trillion in 2025. Those numbers point to a market large enough to justify platform-level bets, especially for enterprise software vendors that already control collaboration and workflow entry points.
What changes next
The near-term test will be adoption: how quickly enterprises move from pilots to production workflows and whether Wukong’s agent layer proves reliable for real business operations. If the CLI-based control model and skills ecosystem can handle complex enterprise integrations, Wukong could become a template for B2B AI across China. If early pilots show clear ROI, enterprises may expand agents from internal admin tasks to customer-facing operations. The next signals to watch are customer references, integration depth across Alibaba’s broader ToB ecosystem, and whether competitors respond with their own enterprise agent platforms.
Sources
- DingTalk official site (AI DingTalk 2.0 / Wukong entry)
https://www.dingtalk.com/ - Securities Times (STCN) — “Alibaba releases the world’s first enterprise Agent platform ‘Wukong’”
https://www.stcn.com/article/detail/3680679.html - 21st Century Business Herald — “Alibaba launches Wukong platform”
https://www.21jingji.com/article/20260317/herald/0a1d0919a17a2e7b3828d68f9af54af2.html - 36Kr — “Alibaba releases enterprise AI work platform Wukong”
https://www.36kr.com/p/3726501820807554 - CAICT — “AI Industry Development Research Report (2025)” (PDF)
https://www.caict.ac.cn/kxyj/qwfb/bps/202602/P020260202487301304903.pdf
