Following the integration of its Doubao assistant into mobile phones, DingTalk has now unveiled its "Project D," bringing the future of AI handling work into clear view.
Recently, DingTalk launched the enterprise-exclusive hardware DingTalk Real and an AI system called Agent OS. CEO Chen Hang stated that in the future, all AI Agents on DingTalk will be built and operated based on Agent OS, enabling AI to connect directly with the physical world.
DingTalk harbors significant ambitions; it aims to lead its ecosystem partners away from the application architecture of the mobile internet era by creating an operating system for the AI age, making the concept of "silicon-based employees" a reality.
At this year's OpenAI developer conference, Sam Altman predicted that the first truly meaningful Agents would enter corporate structures, becoming a "new type of worker" involved in production. Human roles would be compressed to decision-making and issuing commands, while the execution layer would be handed over to machines.
However, a simple chat window is far from sufficient to carry such an ambition. To turn AI's capabilities into productive power, a unified AI system capable of coordinated scheduling is needed, one that can interconnect fragmented models and platforms to genuinely liberate productivity.
As a crucial B2B entry point in Alibaba's AI strategy, DingTalk aims to consolidate AI capabilities at this critical juncture, aspiring to become the Android or iOS of the artificial intelligence era and truly "monetize" AI's potential.
The year 2025 marks an explosion of AI applications; while everyone recognizes the AI wave has arrived, the landscape remains chaotic and unclear.
On one hand, AI capabilities are unprecedentedly powerful, capable of writing reports, creating videos, programming, and making predictions. Yet, the proliferation of diverse AI applications leaves enterprises scrambling between different AI platforms. Furthermore, AI tools often operate in silos, with disconnected data and unaligned intentions, and individual Agents can typically only function within specific platforms.
This "island state" bears a striking resemblance to the early PC era when software applications were largely incompatible with each other.
A truly intelligent productivity tool should function like this: a user simply says, "Help me prepare for next week's presentation to the boss," and the AI automatically retrieves project data, generates a PPT, books a meeting room, and notifies attendees—all without the user needing to lift a finger.
But this requires a unified command center, not just more independent tools. Clearly, the AI era also needs an ecosystem platform that integrates Agents, enabling them to collaborate and produce together.
As the entity overseeing productivity in the B2B sector, DingTalk has made the first move, firing the starting shot.
On December 23rd, DingTalk released over twenty AI-native products in one go. The centerpiece of this launch is a new intelligent work infrastructure with "Agent OS" at its core and "DingTalk Real" as its terminal. This system allows large models, AI tools, and workflows to operate, schedule, and collaborate uniformly within Agent OS, much like Apps, freeing humans from大量 repetitive procedural tasks.
Through the coordination of Agent OS, DingTalk has successfully implemented multiple high-frequency, reusable enterprise-level Agent scenarios, covering core business processes such as AI travel, AI recruitment, and AI customer service. It can be said that DingTalk is no longer just making minor tweaks along the lines of an office software product but has begun building itself as an entire system designed for AI.
All these functionalities are designed to serve real productivity needs.
Previously, Chen Hang observed that the entire industry, including DingTalk's own team at the time, understood AI merely as a tool—using it to write weekly reports, make PPTs, or generate images and copy. But this, he believed, fell far short of touching the essence. He asserted that work methods must change in the AI era.
"Previously, humans were the primary agents of execution and thought, with AI merely assisting; the essence of the new-generation DingTalk is to make AI the primary agent, endowed with近乎infinite memory and global computational power, capable of active perception, judgment, planning, and execution. Humans then take a step back, becoming decision-makers and commanders, making key judgments based on the results produced by AI," Chen Hang explained.
In essence, DingTalk aims to upgrade AI from "loose capabilities" to "systemic productivity." If Windows once made computers accessible to everyone, then Agent OS aims to make systematic AI accessible to every enterprise.
However, merely connecting Agents is not enough. AI also requires accumulated data, and hardware serves as the entry point for it to acquire data and connect with the real world.
This is where DingTalk Real enters the stage, described by DingTalk as the crucial physical extension of Agent OS, akin to AI's body. "The mobile phones and computers we commonly use are designed for people; today, the DingTalk Real we are creating is a device specifically designed for Agents," Chen Hang stated.
In DingTalk's vision, it can allow DingTalk to work continuously for you on your computer, accessing all your data within the internal environment, solving problems and discovering opportunities for you 24/7 without interruption. Once an enterprise deploys DingTalk Real on-premise, employees can remotely invoke Agents via DingTalk from anywhere.
DingTalk intends to build a new moat in the B2B market through this "integrated software and hardware" strategy. Pure software SaaS services are easily replaceable, but once a business deploys DingTalk Real, deeply integrating its business workflows and physical devices with AgentOS, the migration cost increases exponentially. This creates a physical barrier more robust than mere "user habit."
DingTalk has now formally bid farewell to the application form of the mobile internet era. Chen Hang declared, "Today we can announce that DingTalk has completely transformed, transitioning into an AI operating system."
Many might ask: Why would DingTalk, a super App with 700 million users, take the enormous risk of venturing into hardware, developing an OS, and undergoing such a thorough reconstruction? The answer is simple: because in the AI era, you can't find new continents using old maps. And in the battle for entry points among tech giants, DingTalk must hold the fort for Alibaba in the B2B arena.
Public data shows that in 2025, China's collaborative office software market surpassed 30 billion yuan. DingTalk, WeCom, and Feishu collectively hold over 75% of the market share. DingTalk leads with approximately 32.7%, followed by WeCom at around 23.4%, and Feishu in third place with about 18.9%.
It's evident that DingTalk's former dominant position has been eroded. Notably, even AI startups like MiniMax, Moonshot AI, and Zhipu AI, which Alibaba had invested in, have reportedly switched to Feishu. "Feishu's multi-dimensional tables are at least 12 months ahead of DingTalk," Feishu CEO Xie Xin stated bluntly in an interview this past July.
DingTalk reached a point where radical change was necessary for survival. Against the backdrop of Alibaba's overarching AI strategy, a turnaround became possible.
For the past two years, Alibaba has championed the slogan "Let AI permeate thousands of industries," aiming to redo all businesses with AI and redefine production efficiency, much like the advent of the internet did.
Alibaba CEO Eddie Wu Yongming has internally expressed that the greatest value created by the mobile internet was the reallocation of human attention economy, but this value has an upper limit—everyone only has 24 hours a day. In the AI era, the biggest difference lies in the巨大enhancement of human productivity.
With the tailwind of AI, there is potential to move from simple Q&A to actually getting work done. One can imagine that DingTalk has the opportunity to uncover potential within a办公scenario encompassing 700 million users, becoming another ticket for Alibaba's journey into the AI era. A DingTalk that reaches the capillaries of the实体industry can find more avenues for growth and commercialization, becoming an ubiquitous "screw" in every link of the market's supply side.
Therefore, upon Chen Hang's return to DingTalk, he chose an approach akin to "redoing everything" to reclaim lost ground.
This is no easy task. While AI can permeate the C端consumer market through dialog boxes, the B2B market is comparatively harsher. Organizations do not pay for "intelligence" itself; they pay for "deterministic results." The ability to enhance productivity is the paramount concern.
Consequently, DingTalk's greatest value lies in aligning with the theme of the era: enhancing the productivity of its users and enterprise clients, empowering users to become "super individuals." The new DingTalk uses AI to reallocate time and attention, allowing people to focus their energy on critical decisions and high-value outputs, thereby creating new space in work rhythms.
"Once AI has access to stronger data and environments, software will ultimately fade into the background. People will only need to receive the best delivered results," Chen Hang asserted. "Whoever can first help AI agents understand industries and enterprises will inevitably emerge as a dark horse in this productivity revolution."
Initial results under the new strategy are emerging. By December, daily AI calls on DingTalk had increased fivefold over the year, and the total number of Agents created by enterprises and developers on the platform had tripled annually.
"Before August, only 10% of people believed in DingTalk's AI transformation; now, 30% believe," Chen Hang described, likening it to a snowball effect: those who choose to believe first start running, adopting DingTalk's new AI products and methods to get ahead.
DingTalk cannot afford to lose this AI battle. If successful, it will redefine industry delivery standards, transforming DingTalk from an App into an indispensable "infrastructure" within corporate offices, thereby solidifying Alibaba's dominant position in the B2B market.
Using AI to rebuild an Alibaba serves as an anchor for re-evaluating the value of this giant corporation, with DingTalk being a critical link for commercialization and scenario implementation. This elimination race has only just begun, and DingTalk must continue to accelerate its pace.