When the titans clash, the pawns go first.

A battle for the office workers' desktop

In 2026, AI office agents have witnessed explosive growth. According to data from Analysys' *China Office Agent Platform Market Insight Report for Q2 2026*, the total monthly visits of 17 mainstream domestic desktop AI-native office agents exceeded 60 million in June; compared with the 20 million visits in March of the same year, the scale tripled in just three months. Among them, Tencent's WorkBuddy stands out far ahead, with monthly visits of a single product reaching 20.97 million, which has surpassed the sum of the visits of the second-ranked TRAE IDE and the third-ranked product, firmly taking the top spot in the industry.

Behind the surge in traffic, leading domestic internet giants have made intensive adjustments to their organizational structures and gone all-in on the AI office track. Tencent, Alibaba, ByteDance, Baidu, and Kingsoft have successively launched their products, marking the official end of the internal project horse-racing model in the industry and the start of fierce competition with concentrated resources. When the commercialization of general large models in the C-end consumer market cools down, computer desktops and workplace workbenches have become the main battlefield with the greatest monetization potential for large models.

01

What is happening in AI office?

Over the past two years, the domestic AI office track has long been in a trial-and-error state of "flourishing in all its forms". Multiple business lines within major giants have carried out R&D in parallel, leading to serious homogenization of product functions, scattered scenarios, and internal resource consumption. Most AI tools only stay at the shallow auxiliary level of "document polishing and PPT generation", and are not truly embedded in the core office processes.

However, three organizational reforms in July 2026 completely ended the savage growth stage of the industry.

Industry comments from Yicai Global clearly point out that the competition in the AI office track has "shifted from horse-racing to integration", and the industry has officially entered a mature stage of fierce competition among giants, with the core goal shifting from "function trial" to "underlying reconstruction".

Within one month, Tencent, Alibaba, and ByteDance completed organizational restructuring at the same time, shifting from "internal horse-racing" to "concentrating forces" — this is no coincidence, but a strategic consensus.

On July 2, Alibaba disclosed its integration plan, merging three Agent products QoderWork, Wukong, and MuleRun into "Qianwen Office", and launched a small-scale test on July 27.

In late July, Tencent merged the QClaw-related business and team into the WorkBuddy system.

On July 30, ByteDance announced via internal email that the Feishu product team and the Doubao product team were integrated, and Feishu head Xie Xin reports to Doubao head Zhao Qi.

It is clear that this is not a trivial move to add an AI assistant to Word, but a direct bottom-up approach to reconstruct the underlying logic of the entire office software.

02

Why are all major manufacturers chasing AI office?

The collective all-in on AI office by giants is not a chase for short-term trends, but an inevitable choice forced by the dilemma of industry commercialization, as well as a strategic bet on the enterprise service track in the next ten years.

Over the past three years, generative AI large models have been continuously implemented and iterated, but C-end products for ordinary consumers are generally trapped in profit difficulties. High computing power costs, homogenized dialogue tools, and the pain point of low-frequency usage by users make it difficult for pure chat AI products to form a closed commercial loop.

Office scenarios are daily rigid needs for workplace people, with high frequency, strong willingness to pay, and a huge enterprise payment market, which naturally becomes the preferred track for the implementation of large models.

Data from iiMedia Research shows that the scale of China's AI agent market reached 80.4 billion yuan in 2025, a year-on-year increase of 123.2%, and is expected to reach 696.8 billion yuan in 2030. Globally, according to estimates by market research institution 360iResearch, the global AI office software market was valued at 31.98 billion US dollars in 2025, is expected to increase to 35.49 billion US dollars in 2026, and is expected to reach 71.27 billion US dollars by 2032.

From a deeper perspective, the core of this competition is the fight for the next-generation internet traffic and service entrance. In the PC era, browsers and desktop software are the core traffic entrances; in the mobile internet era, WeChat and the APP matrix control user traffic; in the AI era, the first desktop agent after the user turns on the computer will become a new traffic and data hub.

Office scenarios carry all-dimensional core data such as enterprise approval, team collaboration, document data, project management, and schedule scheduling, and are the core carrier of personal productivity and enterprise operation efficiency.

Whoever seizes the AI office entrance will have the say in enterprise services, and can later extend to diversified commercial forms such as customized SaaS services, enterprise knowledge bases, computing power leasing, and industry solutions, unlocking ten years of industrial dividends. This is also the core reason why major giants do not hesitate to restructure their organizational structures and concentrate core resources to tackle key problems.

03

A battle centered on "office workers' desktops"

Under the same strategic consensus, Tencent, Alibaba, and ByteDance have taken completely differentiated breakthrough paths in AI office based on their own core foundations, avoiding head-on red ocean competition and forming an industrial pattern with respective barriers.

1. Tencent's strongest moat is undoubtedly the trump card of the "WeChat ecosystem"

Tencent's core advantage lies in its ecological barrier, focusing on "personal lightweight and cross-terminal integration". Relying on the national-level ecosystem of WeChat and WeCom, WorkBuddy has built a unique "chat is office" mode.

Users do not need complex operations, only by sending text instructions on the mobile phone, the AI agent on the computer can automatically complete a full set of office operations such as document editing, data sorting, file retrieval, and task summary, and automatically push the results to the mobile terminal after completion, realizing a closed loop of office anytime and anywhere. This cross-terminal linked lightweight experience accurately adapts to the fragmented office needs of a large number of workplace people, and is also the core moat that distinguishes Tencent from Alibaba and ByteDance, which is difficult to be quickly replicated by competitors.

2. Alibaba's strategy is "B-end as the shield"

In the past six months, Alibaba has run three lines internally at the same time: QoderWork does local operations on the desktop, Wukong is tied to DingTalk for enterprise collaboration, and MuleRun does cross-platform task automation. Each has its own team and its own direction, and now they are integrated into one force.

After giving up the red ocean competition of C-end personal tools, Qianwen Office is precisely positioned in the government, enterprise, and large and medium-sized enterprise markets, deeply embedding AI capabilities into core work processes such as enterprise approval, attendance, meetings, project collaboration, and document management.

3. ByteDance empowers both technology and scenarios

ByteDance focuses on the adaptability of small and medium-sized enterprises and internet teams. Through the integration of the Feishu and Doubao teams, ByteDance realizes the deep integration of large model technology and collaborative office scenarios. The generative AI, intelligent task decomposition, and semantic analysis capabilities of Doubao fully empower the core functions of Feishu such as documents, meetings, multidimensional tables, and knowledge bases, realizing full-scenario empowerment such as automatic meeting minutes, intelligent document review, automatic work task decomposition, and intelligent project progress summary. At the same time, relying on the computing power and enterprise service capabilities of Volcano Engine, it focuses on lightweight, high-efficiency, and intelligent collaborative office solutions, which accurately adapt to the office needs of internet and technology-based small and medium-sized enterprises.

Although the three giants each hold their own tracks and build their own barriers, the industry competition in AI office has been upgraded from a single function competition to an all-round comprehensive game of ecology, technology, scenarios, and commercialization. However, homogenized involution is becoming more and more serious, the basic functions of each manufacturer are highly overlapping, and the differentiation only stays at the level of ecology and scenarios, and no irreplaceable core technical barrier has been formed yet.

04

Who is paying for AI office?

Compared with the long decision-making, bidding, and implementation cycles of traditional enterprise services, the commercial implementation speed of AI office far exceeds industry expectations, and the core group supporting the initial commercialization of the track is the massive number of workplace office workers.

At present, the three leading manufacturers have all launched standardized subscription payment systems, starting large-scale monetization. According to the official public pricing information of each product:

Doubao Professional Edition is priced at 68-500 yuan per month, covering layered rights such as basic document processing, advanced AI generation, and intelligent task execution;

Tencent WorkBuddy is priced at 99-999 yuan per month, with multiple versions for individuals, professionals, and enterprises to adapt to different user needs;

Alibaba Qianwen Office is priced at 98-198 yuan per month, focusing on the lightweight usage scenarios of individual workplace people and small teams.

The industry presents a distinct commercial feature of "individuals first, enterprises later". At a time when most enterprises have not yet completed the procurement, deployment, and implementation of AI office, workplace people actively pay for AI office tools to improve their personal work efficiency and reduce repetitive work, becoming the first batch of paying users in the industry. This also makes AI office a rare track driven by C-end individual users to reverse the upgrading of the B-end industry.

But at the same time, "salary has not increased, but subscription comes first" has also become a common pain point for workplace people.

AI technology self-media Houchangcun believes that AI office tools have shifted from free trial to paid subscription, and the industry's traffic dividend is gradually turning into a payment dividend, entering the stage of refined commercial operation. What enterprises ignore is that it is difficult for individual users to develop the habit of paying, and "paying to go to work" has also become the biggest problem on the road to commercialization.

Not only that, as ordinary office workers, will you really pay for the so-called improved work efficiency that actually increases your workload?

05

Conclusion

The essence of giants flocking to AI office is not chasing short-term traffic trends, but the super entrance of enterprise services in the next ten years, and the core say in the productivity revolution in the AI era. The current AI office can already replace more than 80% of repetitive office work, greatly reducing the basic work pressure of workplace people. But the industry has always been unable to avoid a core proposition: AI can efficiently complete trivial work, but cannot replace high-level workplace capabilities such as core decision-making, innovative thinking, and resource coordination.

This office revolution led by giants, paid by office workers, and driven by technology, is completely reshaping workplace productivity and industrial pattern.

In the future, AI will no longer be an auxiliary tool for office, but an agent integrated into the whole process of work, and the real workplace competition will shift from "who is more capable of doing work" to "who is better at using AI to do work".

Finally, leave a question: Can AI really finish all the work for you?

This article is from the WeChat official account "Houchangcun", author: Wen Ruyan, published with authorization from 36Kr.