Summary of the Core Content
This article focuses on a 75,000-word essay written by a DingTalk employee titled "Inside DingTalk," which reveals the real challenges in the implementation of AI products and the deeper issues within large corporate management: AI products may seem cutting-edge but could actually represent "pseudo-demand" (for example, DingTalk's ONE interface fails to address users' core pain points). Large companies use high-pressure, dysfunctional processes to exhaust their employees; managers often override genuine user feedback with their own will. The so-called "AI Native organizations" (those claiming to restructure work using AI) are merely masking their shortcomings under several pretenses—rebranding interfaces as revolutionary, pretending to be technologically neutral, using agile methods to convey managerial directives, and using efficiency improvements to cover up the stress on employees. The article concludes that a true AI Native organization should treat people as individuals, not just tools.
Detailed Analysis
1. The Disguises of AI Native Organizations: Does Replacing an Interface with "Intelligence" Mean a Workflow Revolution?
Many companies claim to be AI Native organizations by creating a unified AI interface that integrates messages, schedules, documents, and approval processes, along with an AI assistant for communication, labeling it the "future workplace." However, employees' main issues are not the lack of access to these tools but the overwhelming amount of information, unclear priorities, blurred responsibilities, and constant interruptions from managers. Take DingTalk's ONE product as an example: it aims to make it easier for tasks to be assigned to the right people, yet it doesn't even clarify who the users are (employees or managers). Moreover, DingTalk's design inherently favors managers (for instance, the "read" feature allows them to urge employees to respond). Instead of reducing employees' workload, AI merely delivers more requests more precisely—this is no revolution; it's just giving old management software a new, "smart" facade.
2. AI Tools Are Not Neutral: They Actually Help Managers Exercise More Control
Office software (such as DingTalk and WeCom) has never been neutral, and the integration of AI only exacerbates this bias. For example, DingTalk's "read" feature allows managers to see who hasn't read messages, making it easier to assign tasks; employees feel monitored. With AI, managers can quickly identify bottlenecks in projects, and the system automatically highlights unresponsive employees, with weekly reports summarized as potential risks—all of which increase their control while limiting employee autonomy. This is similar to how traditional systems use approval processes (if not approved, no responsibility is assumed); AI tools inherently protect those in power, not ordinary employees.
3. Agile Methods Have Been Perverted: From User-Driven to Manager-Directed
Agile was originally a great approach for the internet industry, allowing teams to quickly respond to user feedback and improve products. However, many large companies have turned agile into a process of submitting tasks to managers—updating products daily, collecting requirements in the morning, and conducting evaluations in the evening. For instance, when ByteDance's senior management used their AI tool to provide feedback, the team immediately prioritized those requests, disrupting the normal development cycle. More extreme cases involve optimizing search results specifically for managers' accounts, ignoring the needs of regular users. This is not agile; it's simply a way of directly conveying managers' ideas to employees.
4. "Efficiency Improvements" Are a Deception: The Saved Time Doesn't Benefit Employees
The promise of AI is often promising: fewer weekly reports, less meetings, and more time for valuable work. In reality, employees working on AI products are working overtime, and those using AI tools still face additional burdens. For example, while AI can automate meeting summaries, the number of meetings hasn't decreased; instead, records become more detailed and accountability is increased. If AI allows you to handle 10 tasks a day, the company might assign you 50 more. The saved time goes to the company, not the employees. AI merely makes organizational control over individuals more precise and stressful.
5. The Illnesses of Large Corporate Structures: Employees Under Pressure Become "Tools" within Systems
The article highlights various issues typical of large corporate structures, such as daily product updates, morning and evening meetings, score-based work evaluations that affect promotions, and confidential projects that increase collaboration costs. The author even mentions experiencing medical treatment due to the high pressure. Employees are treated like mere tools within these systems: tasks are rushed, tolerance for mistakes is low, and personal health is neglected. No amount of AI can change this fundamental issue; such management practices essentially exhaust employees.
In conclusion, the article argues that a true AI Native organization should first address issues related to power, responsibility, and boundaries—using AI to reduce unnecessary interruptions, clearly defining when human decision-making is needed, and holding managers accountable for the design of processes. Unfortunately, such organizations are still rare today.