The emotional 48-hour "dismissal" of Wu Zhao and the rapid appointment of Chen Yusen represent not just a change in CEO at DingTalk, but Alibaba's simultaneous reworking of its approach to people and products in the AI era.

Content/Ode to the Goose

Proofread/Mangfu

What DingTalk needs is not reconstruction, but a thorough disillusionment.

Wu Zhao's (Chen Hang) second departure marks a meaningful pause in Alibaba's ambitious "Northern Expedition" for B-side AI. From DingTalk's rebirth to the journey for knowledge like the Monkey King, over the past 437 days, Wu Zhao has attempted to replicate the entrepreneurial miracle of Lakeside Garden eleven years ago at DingTalk, which now has over 1,900 employees, using almost violent product development rhythms and management methods.

However, when a product manager exposed internal rifts in a 75,000-word essay, when a former vice president who had already left responded in agreement, when Alibaba's Partnership Committee rarely stepped in to define the situation, and when the announcement of DingTalk's CEO replacement was issued across the internet, everything came to an abrupt halt.

Responsibilities were clearly delineated. The issues at DingTalk were not Alibaba's problems but problems with Wu Zhao's management style. This preserved the legitimacy of the group's AI strategy, responded to internal and external doubts, and cleared the moral burden for the next CEO.

The chosen successor is Chen Yusen, the 34-year-old Vice President of Alibaba Cloud Intelligence Group, born in 1992, setting a record for Alibaba's youngest business unit CEO.

Wu Zhao's exit signals the official end of Alibaba's reliance on strong-willed individuals and physical combat in the enterprise AI battlefield. What Chen Yusen inherits is the interrogation faced by Alibaba's AI strategy in this entire era.

Part.1 The Chosen One

Chen Yusen's background easily reminds one of Jiang Fan from years past. Both have technical backgrounds, had their startups acquired by Alibaba, and were quickly entrusted with significant responsibilities.

However, the biggest difference between Chen Yusen and Jiang Fan is that Chen is taking over not a rising Taobao but a DingTalk in crisis.

A month before his appointment, Chen Yusen was still introducing his new product at the Alibaba Cloud Summit as the head of "MuleRun." Clearly, Chen Yusen was not a successor Alibaba had long prepared for DingTalk but was brought in temporarily.

This also illustrates the selection logic: what Alibaba wants at this juncture is not another operations veteran with DingTalk's DNA but a technologist who still carries the DNA of an entrepreneur.

But what is most intriguing about him is his approach to judging AI products. This approach is almost the opposite of Wu Zhao's in every detail.

Over the past year, Wu Zhao-led DingTalk has released dozens of AI products, including AI Note-Taking, AI Search, AI Spreadsheets, ONE, Agent OS, DingTalk Real, Wukong... a proliferation of features.

When developing MuleRun, Chen Yusen took a completely opposite approach. He believed that products like OpenClaw, which require users to rent cloud servers, buy Mac Minis, and install various software, are "only playable by geeks." His exact words were: "We don't even let them rent virtual machines, buy new computers, or download anything."

This is a product philosophy based on ordinary people. The deepest criticism of ONE in "Inside DingTalk" is precisely that it serves not ordinary employees but an imagined high-net-worth manager.

Yousu mentioned a detail in the article: a product employee excitedly reported to Wu Zhao that Country Garden wanted to use ONE cards to dynamically assign tasks to security guards and cleaners. Wu Zhao did not appreciate this case; he believed ONE was not meant to serve security guards and cleaners but bosses, managers, and high-net-worth individuals.

This small moment reveals a fundamental difference in product philosophy. Chen Yusen believes in lowering barriers to the point of having none, while Wu Zhao believes products should target those with the highest willingness to pay.

Besides their understanding of user barriers, Chen Yusen and Wu Zhao also have vastly different philosophies regarding the relationship between technology and processes.

Wu Zhao said, "Break DingTalk apart and rebuild it with AI to create Wukong." When such grand narratives are applied at the product level, they often result in AI for AI's sake, cramming AI into every aspect and ignoring users' existing workflows.

Chen Yusen's methodology is almost the mirror image. MuleRun "doesn't reinvent Word, doesn't reinvent PowerPoint" but embeds itself into the software and processes users already employ. "You work the way you always have, and AI handles the trouble (troublesome tasks) behind the scenes."

This restraint directly responds to the criticism of ONE in "Inside DingTalk." Yousu wrote, "Good design for AI work products is not about having the system detect work earlier but about allowing users to better master their work."

A statement Chen Yusen made in an interview almost echoes this sentiment: "Large models are glue, not the main body."

Their differing understandings of "letting go" may be the deepest divergence between the two.

Wu Zhao believes management is about visibility—read receipts, late-night checks, seeing who's still in the office, Python exams... The underlying logic of all these methods is gaining control by making people visible.

Chen Yusen's methodology is the opposite. During MuleRun's development, he described a memorable scene where, after discussing requirements with AI, "everyone went home, took time off, and then a day or two later, AI had completed all the features and finished bug testing, almost ready for launch."

He repeatedly emphasizes one sentence: "Never think you're better than AI."

Placed in the context of DingTalk, this statement is almost revolutionary. After Wu Zhao's return, he set a rule for the team that "over 30% of the code must be generated by AI, and no documentation should be written." On the surface, this appears to embrace AI, but it essentially remains a top-down mandatory rule.

Chen Yusen's approach to "letting go" is more like a product philosophy. True AI-native means allowing processes to proceed without constant human oversight and trusting the results.

These two logics essentially represent work philosophies from two different eras.

Part.2 Why Now, Why Him

Just 48 hours before Chen Yusen took over, Alibaba announced the establishment of the Token Foundry business unit, directly overseen by Group CEO Wu Yongming.

This marks Alibaba's third AI organizational restructuring in three months, following the establishment of the ATH business group (Alibaba Token Hub) in March, the creation of a new Group Technology Committee and the upgrade of the Tongyi business unit in April, and now another step up in June.

AI has been personally elevated to a CEO-level project by Wu Yongming.

During an earnings call, Wu Yongming explicitly stated: Over the next five years, Alibaba Cloud and AI commercialization annual revenue must surpass $100 billion. This is an aggressive goal, meaning all Alibaba's business lines must either contribute incrementally or at least not lag significantly.

DingTalk is Alibaba's only super traffic portal besides Taobao, with 800 million users and tens of millions of enterprise organizations. In the AI-to-B battlefield, it should be Alibaba's most critical asset.

However, Wu Zhao's actual output over the past 437 days, at least judging by the group's decisive action to cut ties and replace the leader, did not satisfy the group.

The most intuitive (intuitive) contrast comes from Feishu. Feishu CEO Xie Xin publicly stated, "If your office tools are mainly for clocking in rather than creation, then a year's accumulation might be 10 million clock-in records, and AI can probably help you predict which employees will be late tomorrow."

This remark is harsh but precise. It points to the foundational issue with DingTalk's AI: when a product's core data over the past decade has been attendance + approvals + read receipts, its depth of value that can be excavate (mined) in the AI era is limited.

Commercialization figures also support this judgment. According to Tencent DeepNet, in the first half of the fiscal year ending September 2024, DingTalk's software subscription ARR exceeded $200 million; Feishu predicted an ARR exceeding $300 million for the same year. This means Feishu, with a smaller market share and user base, achieved faster paid growth.

Yet, only 190,000 of DingTalk's 800 million users are paying organizations, indicating a low conversion rate.

This is the most straightforward commercial reason for Alibaba's leadership change, but it's far from the whole story. The real blow that unseated Wu Zhao was not commercial metrics but cultural definition. Looking back at the first internal post by Alibaba's Partnership Committee, the keyword is only four characters: Treat People as People.

Such public characterization is extremely rare in Alibaba's history. It's not just a negation (negation) of a management style but a negation of an entire era's narrative.

This leadership change essentially represents Alibaba's precise public narrative shift.

First, the group's AI strategy is correct, so Wu Yongming, the partnership system, and the overall direction remain intact. Second, DingTalk's AI transformation has issues, but they are defined as management style problems, not strategic direction problems. Finally, responsibility is clearly placed on Wu Zhao personally—an almost obsessive management style incompatible with modern organizations and the AI era.

This shift allows Alibaba to stop the bleeding without altering its course. But the real question it answers is another: When a former hero can no longer solve new era problems, does the group have the courage to replace them?

Alibaba this time gave a definitive answer, faster than anyone expected.

Part.3 Chen Yusen's Real Problem Isn't AI

To pour cold water on Chen Yusen: DingTalk's issues have never been just about whether AI is done well.

Yousu's five-character diagnosis of "greedy and anxious" in "Inside DingTalk" is worth any successor's repeated reading.

She describes the motivational structure behind ONE: reducing user burden, upgrading DingTalk, unifying organizations, and selling tokens for the group. These four goals conflict with no clear priority.

The most fatal tension arises from "serving bosses or serving employees?" and "Is the CEO's preference the product's direction?" These two relationships will not automatically disappear under Chen Yusen; they are structural legacies of DingTalk as a product.

Among DingTalk's 800 million users, the vast majority are not knowledge workers like those on Feishu but price-sensitive organizations in manufacturing, retail, education, and more.

These users initially chose DingTalk for its direct people management; now, transforming it into an AI-native work platform cannot mean abandoning this base. Simultaneously, DingTalk must compete for high-net-worth clients already on Feishu or WeCom.

This is a genuine impossible triangle, and it permeates every business line at Alibaba today.

Qianwen, Taotian, and Alibaba Cloud all face complex goals like C-side entry form shifts, multi-front battles in instant retail and AI, serving internal businesses while maintaining external competitiveness, and more.

DingTalk is merely the sharpest cross-section of this structural contradiction.

What Chen Yusen faces next is not just pushing aside ONE's corpse and growing Wukong. He must answer a more fundamental question: In a group context that wants everything, can he define a relatively clear product boundary for DingTalk?

This tests not technical ability but the ability to say no.

Putting all this together, Chen Yusen faces three narrow gates he must pass through: Can he balance being an entrepreneur and an organizational manager? Can he truly let DingTalk let go? Can he restore morale at DingTalk?

Critiques of Wu Zhao centered on his use of an early-stage entrepreneur's approach to manage a mature organization of 1,900 people at its peak, resulting in a severe mismatch between organizational scale and management granularity.

He would still check who wasn't at their desk late at night, but he no longer faced the seven or eight people crammed in Lakeside Garden during DingTalk's infancy but a complex system spanning Hangzhou, Beijing, Shanghai, and Shenzhen.

Chen Yusen led a small team of twenty to thirty people with MuleRun. Jumping to DingTalk's scale, he faces even harsher tests than Wu Zhao did initially; he can neither revert to MuleRun's flat structure nor inherit Wu Zhao's high-pressure mechanisms.

This is a true narrow gate.

Moreover, Chen Yusen's repeated emphasis in MuleRun on "letting AI handle it" will immediately collide with a wall at DingTalk, whose core value over the past decade has precisely been about not letting go. The most profitable parts of DingTalk essentially sold managers a remedy for their lack of trust.

Now, Chen Yusen must inject "letting go" as a product philosophy into DingTalk without negating its past, redefining its core value. This is an epic-level product transformation.

Restoring morale at DingTalk may be even harder than the first two challenges.

Yousu and Ma Ruila's articles are not fundamentally about products but about people.

At 34, Chen Yusen is younger than many veterans on his team. His advantage lies in the refreshing aura of MuleRun—"discuss with AI and then take time off, only to find the product completed upon return." If DingTalk's organization can absorb this aura, it will be healing in itself.

But this on the premise of (prerequisite) is whether he can withstand the group's KPI pressure and give the team time without constant urge (urging).

Looking back at this leadership change, its significance extends far beyond a personnel adjustment at one company; it reveals a crack in the era.

On one side is the success formula of the old internet era: iron-fisted, aggressive, overtime, burning out, strong-willed, late-night lights, treating work as a career, treating employees as brothers...

This logic swept through China's internet battlefield from 2010-2020, creating countless phenomenal products. Wu Zhao was one of the purest incarnations of this logic.

On the other side is the new formula of the AI era: restraint, letting go, trust, embedding, low barriers, product philosophy, user perspective, treating AI as a collaborator, treating employees as people...

This logic is far from proven successful but at least points more closely to the fundamental question of what AI has truly changed.

Wu Zhao represents the afterglow of the former; Chen Yusen represents the dawn of the latter. Alibaba's leadership change essentially represents a choice between the two logics.

But this choice was not made in a vacuum; it was driven by a series of signals: Yousu's 70,000+ words, Ma Ruila's 20,000+ words, the internal post by the Partnership Committee, Feishu's ARR progress, DingTalk's ongoing employee exodus, the urgency of Token Foundry's establishment...

When an organization is large enough, any seemingly sudden leadership change is preceded by countless silent signals that have been unfolding for a long time.

The five characters "greedy and anxious" are Wu Zhao's final legacy to DingTalk—both a verdict and a warning.

What Chen Yusen inherits is not just a CEO position but all the reflections behind these five characters.

This is the real issue for DingTalk, for Alibaba, and for everyone trying to build products in this AI era.

END