A former DingTalk product manager's viral resignation essay has turned a failed AI rollout into a wider test of Alibaba's management culture. The issue is no longer whether Project ONE worked as promised, but whether the company can build AI products without exhausting the teams behind them.
When a former DingTalk product manager published a resignation essay in early June, the post did not read like a routine exit note. It was a detailed account of Project ONE, DingTalk's marquee AI effort, and it quickly became a proxy for a bigger question now hanging over Alibaba: can one of China's most important technology companies execute on AI without letting internal pressure damage the work?
The essay, published under the title "Inside DingTalk," described a project that began with a clear ambition and then became harder to govern as expectations rose. Project ONE was supposed to make DingTalk feel less like a collection of workplace tools and more like an intelligent operating layer for office work. Messages, approvals, meetings, documents, calendars, and to-do items would be pulled into a single flow, with the system deciding what deserved attention first.
That is a serious product idea. Enterprise software has trained workers to jump between apps all day, and the promise of AI is that some of that friction can finally be reduced. For Alibaba, DingTalk matters because it already sits inside the daily routines of millions of Chinese corporate users. If AI can make that workflow smarter, DingTalk becomes more than a collaboration platform. It becomes distribution for Alibaba's broader enterprise AI strategy.
The resignation essay argued that the reality inside the project did not match that ambition. The author described shifting priorities, unclear resource decisions, and pressure from leadership that made it difficult for teams to experiment properly. That matters because AI products rarely improve through command alone. They need real user feedback, time to test assumptions, and enough trust for teams to admit when a feature is not working.
On June 10, Alibaba's Partnership Committee responded with an internal memo titled "Loyalty and Growth: That's Alibaba Culture." The committee said the management behavior described in the essay was not the direction Alibaba's culture supports, and it emphasized employee passion and creativity over pressure and mechanical execution. For a company whose partnership structure is designed to protect long-term culture and leadership continuity, that intervention was unusually pointed.
The significance is not just that Alibaba acknowledged the controversy. It is that the rebuke came from the top of the institution rather than from a product or communications team. The Partnership Committee is associated with Alibaba's senior governance and leadership culture, so its decision to address a specific business unit's management practices gave the episode a larger meaning. This was not treated as a personnel dispute. It was treated as a warning about how AI work is being managed.
Why DingTalk Makes This Bigger Than One Product
DingTalk is not a side experiment. Alibaba has spent years trying to strengthen its enterprise technology business, and DingTalk gives it a direct channel into companies that may eventually buy more AI tools, cloud services, and automation products. As public company materials and industry coverage have long made clear, DingTalk is one of Alibaba's most visible workplace platforms, which is why a breakdown inside its AI transformation carries consequences beyond the team itself.
Project ONE was meant to show how that platform could evolve. The idea of moving from "people look for tasks" to "tasks look for people" captures the appeal neatly. If the software can understand what matters, rank it, and surface it at the right moment, it changes how enterprise work gets done. But that promise also raises the execution bar. The product has to be useful enough to earn trust, accurate enough to avoid creating new noise, and flexible enough to fit different companies' workflows.
The former employee's account suggested that those demands collided with a management style built around urgency. That is a familiar problem in AI right now. Companies want the market to see speed, but the work itself often requires patience. The harder leadership pushes for visible progress, the easier it becomes to reward output that looks impressive before the underlying product is ready.
A second public essay from Ma Ruila, a DingTalk vice president, added to the pressure by reflecting on his own departure and echoing concerns about the organization. Two public accounts from people close to the transformation made it harder for Alibaba to frame the issue as one unhappy employee's complaint. It began to look like a cultural signal from inside a strategically important team.
The Management Question Behind China's AI Race
The episode also lands at a moment when China's largest technology companies are racing to turn AI investment into products that businesses will actually use. Alibaba has Qwen models, cloud infrastructure, and a large enterprise customer base. ByteDance, Tencent, Baidu, and others are chasing the same market from different angles. The competition is real, and so is the pressure to ship.
But AI does not erase the old rules of product management. It makes some of them more important. Teams still need stable priorities. Engineers and product managers still need room to test and revise. Leaders still need to know the difference between urgency and churn. When those basics break down, more compute and bigger model claims do not solve the problem.
That is why the Partnership Committee's memo matters. Alibaba appears to be drawing a line between ambition and the management behavior required to sustain it. The next test is whether that message changes incentives inside DingTalk or simply becomes another internal slogan. If teams still feel punished for uncertainty, the company will struggle to build products in a field where uncertainty is part of the work.
What happens over the next few months will say more than the memo itself. If DingTalk stabilizes its AI roadmap and keeps talent focused, Alibaba can turn an embarrassing public dispute into a useful correction. If more departures follow, the story will point to a deeper weakness. In enterprise AI, the companies that win will not only be the ones with strong models. They will be the ones that know how to manage the people trying to turn those models into products customers can use.
Also read: Microsoft cuts Claude Code from its engineers because the bill came due before the budget year did • A federal court's finding that AI is not required for search reshapes the competitive battlefield for Google and its would-be disruptors • China used ChatGPT to manufacture American opposition to data centers and OpenAI documented exactly how it worked
