Summary of Key Points
In July, the three major office software solutions—WeCom, DingTalk, and Lark—intensively launched upgrades to their AI capabilities: WeCom tested its AI assistant “Dayuan,” DingTalk integrated three AI product lines to launch “Qianwen Office,” and Lark emphasized “multi-agent collaboration.” Each company has its unique approach (WeCom focuses on private domains, DingTalk aims for full coverage, and Lark integrates data). However, these three solutions already dominate 92% of the market, making it difficult for functional upgrades to change the current landscape. The critical question is whether AI in office applications can evolve from merely helping employees with tasks (such as writing weekly reports or summarizing information) to genuinely contributing to business growth (by driving sales or improving performance). As of now, none of the companies have found a clear answer to this challenge.
I. The Unique Features of Each Company’s AI Products and Their Strategic Intentions
1. WeCom “Dayuan”: An Intelligent Assistant in Private Domains
Can be activated with a swipe left; its main focus is on integrating into daily work flows. It can automatically summarize conversations and documents to create weekly reports, extract key points from group chats, generate to-do lists, and analyze WeChat customer interactions (to identify high-potential clients and create customer profiles). Its advantage lies in its access to a private domain of 14 million companies and 750 million WeChat users. However, the features are currently basic (test users have mentioned that it cannot handle complex cross-system processes), and the system is relatively closed-off.
2. DingTalk “Qianwen Office”: Aiming to Be the “Operating System” for Corporate AI
It combines three intelligent agents, including QoderWork and Wukong, and may eventually be built into DingTalk itself. Its goal is to achieve full coverage—helping not only with tasks but also integrating with internal corporate systems (such as manufacturing processes or government approvals). This approach suits medium to large enterprises, but the abundance of features can be overwhelming for smaller companies.
3. Lark “aily”: An AI Colleague That Lives within Data
The upgraded version can work 24/7 on tasks like analyzing spreadsheet data and writing reports, and it supports multi-agent collaboration. Its most notable feature is the “multi-dimensional spreadsheet agent”—users can turn spreadsheets into intelligent assistants with a few clicks; by @-ing it in group chats, it can automatically update progress and analyze data. The advantage is its ability to integrate data effectively, but it has the smallest user base (about 30 million monthly active users), and its target customers are often from sensitive industries like the internet or creative fields.
II. Underlying Logic: Different Approaches Stem from Unique Origins
The different AI strategies of the three companies reflect their fundamental differences:
- WeCom: Focused on Private Domains
WeCom’s primary purpose is not to improve office efficiency but to connect with WeChat users. Therefore, “Dayuan” focuses on customer-related tasks, such as assisting sales in understanding needs and identifying conversion opportunities, with the goal of helping companies generate revenue rather than just saving costs.
- DingTalk: Full Coverage + Strong Control
With 200 million monthly active users and coverage across various industries (manufacturing, government, education), DingTalk aims to create a comprehensive AI system that incorporates its 11 years of accumulated organizational structures and approval processes, serving as the foundation for corporate AI transformation.
- Lark: Emphasizing Data Integration
Since most of its users are from internet-related fields, the main issue is the dispersion of information (group chats, documents, spreadsheets being managed separately). Lark’s strategy is to make intelligent agents part of the data flow, turning them into active collaborators that can help with tasks.
III. The Current Market Landscape and Its Challenges
The three companies together hold 92% of the market (DingTalk leads with 200 million monthly active users, followed by WeCom with 100 million, and Lark with 30 million). Functional upgrades are unlikely to change this distribution. However, each company faces significant challenges:
- WeCom: Data Privacy Risks
Relying on WeChat chat records and customer data, any breaches or misuse could lead to loss of user trust and regulatory issues.
- DingTalk: Features Are Too Complex for Small Companies
An AI system that requires two weeks of training may be too burdensome for small businesses, potentially resulting in limited adoption.
- Lark: Limited User Base and Sensitive Customer Segments
Its smaller user base, mainly from internet-related industries, makes it more vulnerable to economic fluctuations, which could lead to customer churn.
- Competitive Threats from Generalized Large Models
AI models like DouBao and Kimi, although not designed for office software, are becoming increasingly capable of handling conversations and information. If workers can use these models to complete tasks across applications, the barriers created by these companies may be overcome.
IV. The Ultimate Question: Does AI in Office Applications Aim to Reduce Burden or Generate Revenue?
Currently, most AI features focus on reducing burdens (writing reports, summarizing information, automating tasks), which save time but do not directly contribute to revenue. The real goal is to generate income—helping sales close deals, improving factory efficiency, or attracting customers for stores. Achieving this requires a deep understanding of industry-specific processes (such as manufacturing or retail). All three companies are experimenting, but no clear path has been found yet. WeCom is exploring private domains, DingTalk is testing industry-specific applications, and Lark is focusing on data integration. The article concludes that “no company has succeeded yet”; the one that first demonstrates how AI can truly generate revenue for businesses will win the next round of competition.
Conclusion
This battle for AI in office applications is about more than just functional upgrades; it’s about finding new growth opportunities. While the existing market is stable, AI could open up new possibilities—not by competing for existing users but by creating value for businesses. The ultimate winner will be the one that transforms from a tool provider to a business enabler.