DingTalk Launches Agent OS, the World’s First AI-Native Operating System for Work
Want to read in a language you're more familiar with?
DingTalk has unveiled Agent OS, positioning it as the world’s first AI-native operating system built specifically for intelligent work collaboration.
December 23 — At the AI DingTalk 1.1 “Mulan” product launch and ecosystem conference, DingTalk introduced Agent OS, which it describes as the world’s first work-oriented intelligent operating system purpose-built for AI. The release marks the beginning of a new era of human–AI collaboration in the workplace.
Chen Hang, Founder and CEO of DingTalk, explained that “Mulan” symbolizes the early dawn of the AI collaboration era. Going forward, all AI Agents within DingTalk will be built and operated on Agent OS, enabling AI systems to connect directly with the physical world. At its core, Agent OS is designed around multi-agent collaboration, providing a unified and secure framework for AI execution in complex enterprise environments.
The system is supported by a comprehensive collaboration architecture, including the DingTalk ONE unified interface, dedicated AI hardware DingTalk Real (described as the “physical body of AI”), AI Search & Ask, the general-purpose task-processing Agent Wukong, and the enterprise AI platform DEAP. Together, these components ensure AI agents can operate safely, efficiently, and at scale within organizations.
More than 20 AI-powered products were also unveiled at the event. In manufacturing, Order and Quality Agents can convert order images into production schedules and proactively flag potential failures. The AI Travel Assistant can generate full travel plans in under one minute while cutting costs by up to 15%, while AI Customer Service solutions enable rapid cold-start deployment and continuous self-improvement. AI DingTalk 1.1 also upgrades core features such as AI Search and AI Sheets, and introduces new capabilities including face-to-face translation.
Chen emphasized that Agent OS is an open ecosystem, with DingTalk working alongside partners to co-develop industry-specific models and solutions. The goal is to foster an enterprise AI ecosystem characterized by “thousands of intelligent agents, flourishing as one.”
