An extensive resignation letter, exceeding 70,000 characters, has gone viral online, thrusting China's premier office software, DingTalk, into a maelstrom and unveiling its internal and external predicaments.

Internally, the high-pressure management style has fueled intense internal friction and exacerbated talent attrition. The flagship project 'ONE,' which bears significant AI ambitions, struggles amidst conflicting KPIs. Externally, competitors Lark and WeCom are closing in, exerting pressure on both user experience and commercial expansion. DingTalk finds itself at a pivotal crossroads where it cannot afford any missteps.

The turmoil engulfing DingTalk transcends mere personnel and product issues within a single company; it illuminates the common challenges of reshaping organizational relationships and product values as major internet firms transition into the AI era.

DingTalk's 'Divergence': AI Empowerment or Misstep?

A resignation article titled 'Inside DingTalk,' spanning approximately 75,000 characters, has sparked significant controversy both on Alibaba's internal platform and across the broader internet. This piece not only represents the voice of a departing employee but also serves as a profound reflection on DingTalk and the entire workplace ecosystem.

The genesis of DingTalk's 'divergence' did not stem from external competitive pressures but rather from the collapse of internal values within the team.

In internal campaigns such as 'GaiYuan' and 'WangShu Action,' mentioned in the resignation letter, DingTalk's product team deviated significantly from the essence of product iteration and upgrades. Research and development efforts, which should have focused on understanding user needs, improving product usability, and enhancing work efficiency, were largely diverted towards formalism and cutthroat competition.

For instance, the so-called 'WangShu Action' was an absurd practice: Instead of focusing on improving their own product, DingTalk's product team would monitor when the lights went out in the opposing Lark building to gauge their workload and, in turn, push their own team to work overtime.

Over time, an unspoken belief took root: Overtime became synonymous with work attitude, intense internal competition with loyalty, while genuine product innovation was marginalized.

More alarmingly, DingTalk's once-touted AI product, 'ONE,' achieved 3 million DAU in its first week but quickly witnessed a plunge in user retention, ultimately fading into obscurity. This was not due to technical shortcomings but because its implementation triggered 'read anxiety' and a sense of constant surveillance.

In ONE's design, work messages were pushed to users as short video cards. Users only needed to glance at the AI-generated summaries on these cards, and the original IM conversation would be marked as 'read.'

This meant that an employee's status was marked as 'read' even before they personally opened the message. Such a design stripped employees of their autonomy over the speed and manner of information reception, subtly increasing the psychological pressure to respond immediately.

When AI ceases to be merely an auxiliary tool and instead becomes a device for urging and supervising work, even going so far as to turn 'read but unreplied' messages into performance metrics, the original burden-reducing tool transforms into an all-encompassing monitoring device. Managers can check at any time, but employees have no means of escape. This extreme control, lacking any psychological buffer, blurs the boundaries between work and rest, constituting invisible psychological oppression.

The author believes that this is not a technical issue but a matter of the developers' values. It exposes a fatal flaw: If the tool's nature is overly amplified and considered solely from the perspective of managers and payers, it becomes a poison that stifles organizational vitality.

Alibaba Intervenes, DingTalk Seeks Renewal

In response to escalating internal conflicts, the Alibaba Partnership Committee broke its silence in a rare move. In an article titled 'With Compassion, Integrity, and Growth: This Is the True Essence of Alibaba Culture,' posted on the internal network, it stated that DingTalk's current management model 'does not align with Alibaba's culture.'

This intervention not only addressed workplace discipline but also, at a crucial moment in Alibaba's AI transformation, prompted a profound reflection and conscious improvement of the organization's foundations.

Firstly, a course correction: From 'cold KPIs' back to 'compassion and integrity.'

As Alibaba's vanguard in the To B sector, DingTalk shoulders significant commercialization and market expansion pressures. Amid fierce competition, some management practices deviated, focusing solely on efficiency at the expense of humanity, thereby fostering internal discontent.

In response, the internal letter from the Partnership Committee reemphasized Alibaba's cultural essence: 'Mutual respect, treating people as individuals, with compassion and integrity.' This is not merely a slogan but a fundamental requirement for managers.

Secondly, a warning: Innovation in the AI era does not rely on 'high pressure' but on 'human-centricity.'

The internal letter stated, 'Innovation has never relied on high pressure and mechanical execution, especially in the AI era.' This is a far-sighted perspective. Before the advent of large model technologies, truly competitive work was low-end and repetitive; now, it has shifted to the creativity and imagination of top talent.

For Alibaba, the Partnership Committee's direct intervention demonstrates the highest decision-making level's emphasis on organizational health. It serves as a warning to DingTalk and a guarantee to all employees across the group. From an industry standpoint, this is a crucial value guide, breaking away from the past's internal competition management model and redefining the relationships between enterprises and management innovation in the new era.

It sends a clear signal: The old approach of relying on high pressure to achieve results is outdated. Losing core talent before the AI era's second half even begins means the enterprise is already half-defeated.

A Major Test in the Agent Era

Beyond internal adjustments, DingTalk faces even more severe structural challenges.

Previously, everything in DingTalk revolved around 'people.' Meetings required human initiation, approvals needed human submission, documents required human organization, and knowledge bases required manual updates. DingTalk's role was to serve as a framework for managing the three elements of 'people-processes-organization' within a system, enabling visibility, control, and traceability in enterprise management.

However, the emergence and development of AI Agents have altered these long-standing rules and redefined the core processes of enterprise work.

AI Agents are not merely traditional conversational assistants or content generators; they have evolved to possess autonomous perception, proactive behavior, and closed-loop execution capabilities, becoming self-running digital entities. They no longer just 'assist people in operating' but have transformed into 'replacing people in completing' tasks.

Imagine a future work scenario: AI Agents seamlessly integrate with all modules of enterprise communication, approvals, document management, and knowledge base updates. Without rest or supervision, they automatically complete a series of repetitive, procedural tasks such as information aggregation, process submission, data organization, and business reporting.

From daily attendance statistics and reimbursement approvals to complex project document archiving and business data updates, all can be automatically executed by a single intelligent agent without manual intervention. Therefore, the value of software no longer lies in whether it can make people work faster but in whether it can enable seamless Agent invocation and truly replace human labor.

If DingTalk continues to exist solely as a 'control tool' without transforming into an 'AI-native service foundation,' its previously essential collaborative value will be significantly replaced and compressed by AI.

Conclusion

In DingTalk's case, it was not merely a resignation article that ignited public opinion but rather a long-simmering internal conflict that finally found an outlet.

This incident reflects the common challenges faced by major firms during the AI transformation: The inertia of old, strong-control management practices is undermining the new technological dividends. The Alibaba Partnership Committee's 'decisive intervention' not only slammed on the brakes for DingTalk but also pointed it towards a new path of returning to its cultural roots.

However, DingTalk still has a long road ahead in balancing efficiency and humanity, KPIs and creativity. Whether it can completely shed its 'panoptic surveillance' misalignment, reshape a culture of compassion and integrity, and redefine its product value in the AI Agent era will determine whether this national-level application can navigate through cycles.