Reporting by Wu Xun and Ma Shuye, Editing by Liu PeiAlibaba on Thursday announced a management reshuffle: Chen Hang stepped down as DingTalk CEO, with Chen Yusen, 34, taking over. At present, Chen’s desk has already been cleared out, and he has also exited DingTalk’s all-hands group chat.As DingTalk’s founder, Chen had originally been entrusted with high expectations by Alibaba. After he returned in April 2025, he was also seen for a time as Alibaba’s go-to hard-charger to seek a breakthrough in AI for B2B—someone who could fight the tough battles.At last year’s product launch, Chen said he was thrilled by this wave of AI and wanted to rebuild DingTalk from the ground up with AI; he also said this return would be for the next ten years. Back then, almost no one doubted him.It was just unexpected that a little over 400 days later, he would step down quietly for a combination of reasons.A 70,000-Character Post Triggers a ReshuffleNews that Chen was stepping down as DingTalk CEO had already spread internally Wednesday night. Someone posted a “fireworks” sticker in DingTalk’s all-hands group chat of more than 1,400 people, drawing hundreds of likes; after a while, an admin deleted the “fireworks.”But the situation in the chat had already spiraled out of control. More people kept posting “fireworks,” others continued to like them, and the admins kept deleting them.Interestingly, today the news of Chen stepping down sparked heated discussion, yet DingTalk’s thousand-plus-person all-hands group chat was completely silent. “The eye of the storm is calm,” one DingTalk insider explained.As early as three days before he stepped down—on June 7—Chen’s behavior had already started to seem a bit off. It was a Sunday morning, and Chen convened a team check-in meeting. One attendee said Chen had just come back from the group, and although he was usually bursting with energy, he looked visibly tired during the meeting.Two teams presented that day. Chen spoke far less than usual; he no longer pressed on the details and even seemed somewhat distracted.In employees’ eyes, Chen was high-energy and worked with a near self-destructive intensity. But on the other hand, he was known for forcing employees to work overtime, monitoring whether people were at their desks, and using heavy-handed management tactics.Early in his return, Chen kicked off organizational reshuffles and tightened control over employees’ work routines. After that, both voluntary resignations and forced departures were very common. Over a little more than half a year, DingTalk’s headcount also plunged from around 2,000 to just over 1,400 today.Another person at DingTalk told us that after Chen returned, he went into overdrive, and internally they quickly kicked off a new round of AI innovation “horse-race” projects. He asked everyone to proactively claim tasks, form small teams around different directions, and each team usually consisted of just a few people.At the time, there were more than twenty teams. Each team was essentially shut inside a meeting room or office, holding high-intensity discussions about its product ideas and paths to implementation.All 20-plus teams had an equal chance—whoever broke through would get resources, manpower, and financial support. Since the second half of last year, several products emerged, including ONE, Dingtalk A1, and “AI Tingjian” (“AI Hears”). These products also made high-profile appearances at DingTalk launch events for a time, but later fell into controversy.At first, the most watched was ONE, which Chensaw as the entry point for rebuilding DingTalk. At last year’s launch event, Chen put ONE in the most central position of the entire “AI DingTalk,” but ONE later became highly controversial.On June 4, a roughly 75,000-word resignation essay, Inside DingTalk, spread from Alibaba’s internal network to the public internet. The author was Teng Yaxin (nickname “Yousu”), the product manager of DingTalk’s ONE. She reviewed the whole process of ONE—from project initiation, to being pinned with high expectations at the DingTalk 8.0 launch event, to DAU once surging to around 3 million, and then to the entry point being pushed back and ultimately replaced by the new “Wukong” project (an enterprise-grade, AI-native work platform).In early 2026, as strategy tightened, the entire ONE team was merged into the Wukong project, the product was taken offline, and it was no longer operated independently. Yousu attributed this to ONE being bound up with multiple missions, which caused the product to carry far too much information. The underlying reason, she wrote, was that leadership wavered and lacked clarity in positioning. The most obvious example was trying to both serve bosses well and curry favor with employees.The second “ace” product was the recording hardware Dingtalk A1, which for a time was the hottest-selling recording product in China. The eight-person team originally responsible for Dingtalk A1 not only gained support that expanded to 20 people, but also received a marketing budget of over RMB 100 million.However, the AI narrative shifted quickly afterward. Agent products represented by “Longxia” quickly became the new competitive storyline among big tech companies, and the new Wukong project also became DingTalk’s lead role. But the much-anticipated Wukong delivered a so-so experience, and its reputation lagged behind Alibaba Cloud’s coding product Qoder and Alibaba International’s agent Accio Work.Yousu extended the discussion from product failure to organizational management—repeated strategic reversals, product decisions overly dependent on the CEO’s personal preferences, high-pressure iteration, internal team infighting, and more.Her article spread widely and struck a chord in online workplace discourse. On June 8, Ma Ruila, DingTalk’s vice president and head of AI products, posted Outside DingTalk, confirming that he had resigned on May 15. Ma wrote that he left because it was becoming increasingly hard for him to tell whether he was creating products, or merely burning through his body to chase a constantly moving pace.He worked seven days a week, clocking in at 9 a.m. every day, getting home at 2 a.m., sleeping five hours, and then doing it all over again the next day. He’d been chronically sleep-deprived. In response, he said that after finishing Inside the Nail, he felt “heartbroken.” At the end of the article, You Su wrote that the Titanic sank, but the sailors on board could still go look for their next job. Ma Ruila responded: only the sailors who survive can find their next job.On June 10, Alibaba’s Partnership Committee published an internal post titled “With Warmth, Integrity, and Growth—That’s Alibaba Culture,” formally responding to the controversy. It criticized certain DingTalk management practices as “not what Alibaba culture should look like,” and emphasized respecting individuals and protecting creativity. As of today, Alibaba has announced a management reshuffle, and Chenhas stepped down.In any case, Chen Hang’s contributions to DingTalk cannot be easily erased. In its early days, DingTalk started as an enterprise communication tool, and it was able to scale all the way up to become one of the most important B2B products in the Alibaba ecosystem—Chen Hang was a pivotal figure. He has a strong product will and relentless execution ability, and it was precisely this forceful style that helped DingTalk quickly establish a presence in the era of mobile work.But in the AI era, the old logic may no longer work. Alibaba needs fresh blood to rebuild a new DingTalk for the AI era.A Tech GeekChen Yusen taking over as DingTalk’s CEO looks like good news to many. An AI-native organization should be led by young people who understand AI best—this is also the dominant storyline among AI companies today.After taking the role, Chen Yusen will also become Alibaba’s “youngest business unit CEO.” But what truly deserves attention in this appointment is not his age, but an easily overlooked part of his identity.For DingTalk, the successor is a current vice president of Alibaba Cloud Intelligence Group, and his entire growth path has been within the Alibaba Cloud system. He isn’t an outsider parachuted in; he is an insider who has been proven within the system—still holding a VP title—but who also needs to re-earn internal trust.At 22, Chen Yusen founded the cybersecurity company Chaitin Tech, which Alibaba acquired outright in 2019 and continued to run independently after the acquisition. In 2025, he launched a second venture inside Alibaba Cloud and built the AI Agent product MuleRun. From being acquired, to becoming the youngest VP, to incubating a star product internally, Chen Yusen has already become a homegrown AI leading figure within the company.As early as June 2, Chen Yusen’s MuleRun advanced the product toward “enterprise-grade AI collaboration.” Nine days later, he became the new steward of DingTalk—the largest enterprise collaboration platform for Alibaba employees. This can almost be seen as an agent-collaboration paradigm validated in the lab, now being moved directly onto the main battlefield.This reshuffle can be understood as Alibaba Cloud’s methodology being re-imported into DingTalk laterally.DingTalk’s organizational positioning has kept shifting over the past few years. In its early days, it was a collaborative office product that grew relatively independently within Alibaba. In 2019, it was folded into Alibaba Cloud Intelligence; in 2020, Alibaba further pushed “cloud–DingTalk integration,” attempting to turn DingTalk into Alibaba Cloud’s front-end entry point for enterprise customers. That meant the group required DingTalk to make room for cloud computing—this was also why Chenleft DingTalk at the time.By 2023, DingTalk spun back out of the Alibaba Cloud system, returned under the Alibaba Group framework, and emphasized independent development and market-oriented operations.Now, in today’s AI era, DingTalk is no longer just an office collaboration tool. It needs to take on a new role: translating Alibaba Cloud’s technological capabilities in large models, AI agents, and enterprise services into products and scenarios that enterprise customers can actually use. Given Wu Zhao’s personality, it seems hard for him to accept this kind of “supporting role.”So Alibaba Cloud sent to DingTalk Chen Yusen, who had built MuleRun.In media interviews, Chen Yusen looked back on MuleRun’s entire product evolution. After internal testing, he found that running the same prompt in his product and in leading products yielded similar results, because everyone relied on Claude’s API capabilities. Chen believed MuleRun should become a platform that continuously lowers the barrier to professional use.This kind of restraint—standing on the shoulders of giants—is actually rare for someone with a technical-geek background. But it goes far beyond product judgment.“Lowering the barrier” was also a high-priority phrase in Chen Yusen’s conversations. If this methodology is transplanted to DingTalk, it points to a very specific product philosophy: DingTalk’s AI transformation shouldn’t be about piling on features; it should be about enabling ordinary employees to actually use it. As for how much this approach can reshape DingTalk’s long-standing product perception as something “for bosses,” that remains to be seen.It’s also worth noting that the connection between Chen Yusen and DingTalk had already surfaced earlier. In the interview, he mentioned that during internal testing, operations colleagues with zero technical background were already able to use natural language to have an agent automatically operate a browser to do KOL background research, and then sync the results to DingTalk Docs. A series of MuleRun feature updates had, in fact, been aimed at DingTalk-like scenarios.Now, moving from leading MuleRun—a small team of about 50—to DingTalk, a mature organization on a completely different scale, what Chen Yusen has to prove is likely more than just management chops. DingTalk’s commercialization targets were mentioned repeatedly in the two articles “Inside DingTalk” and “Outside DingTalk.” What DingTalk may need is not a revolution, but proof that it can grow.Meanwhile, MuleRun itself has yet to reach product–market fit. A tech executive still waiting to be recognized, armed with a methodology that is still being validated, is stepping into an even more complex fight. That is the biggest highlight of this reshuffle—and also its greatest unknown.更多精彩内容，关注钛媒体微信号（ID：taimeiti），或者下载钛媒体App