Text / Zero Degree

Source / Node Finance

On November 9, Alibaba officially released Qianwen Office 2.0.

One change in this upgrade is particularly noteworthy: it is no longer merely an AI assistant serving individuals, but has begun to support collaboration among multiple people and Agents, allowing AI to be pulled directly into work groups and participate in team tasks.

Just a few days earlier, on November 6, Huawei launched the official version of Xiaoyi Work on the HarmonyOS PC App Store. On October 30, Tencent announced four major upgrades to WorkBuddy Enterprise Edition. In September, Feishu released version 8.0, further strengthening its Agent capabilities.

Putting these dates together reveals an interesting phenomenon: several leading domestic tech companies are coincidentally shifting the focus of AI competition toward office work.

If in 2023 and 2024 big tech competed on large model capabilities, and in 2025 on AI application deployment, then by the second half of 2026, competition is entering a new stage—who can truly take over part of the work?

I have a judgment: in 2027, AI office is very likely to become one of the most important and most intense application battlegrounds for domestic big tech.

Why will AI office become a must-win battleground for big tech in 2027?

First, look at market data.

The "Q3 2026 AI Application Market Development Report" released by QuestMobile shows that as of September 2026, the monthly active user scale of domestic AI office reached 142 million, a year-on-year increase of 50.6%.

The same report shows that in September 2026, WorkBuddy's total active user scale reached 32.353 million, Baidu Dazi 9.77 million, and Qianwen Office 7.905 million.

This set of data at least shows that AI office has passed the early technology trial stage and begun to enter a broader user market.

Of course, these are third-party monitoring calibers, do not represent paid user scale, and cannot be directly equated with the number of enterprise customers.

But the trend is relatively clear: AI is transforming from a chat tool into a work tool.

Why will office scenarios become the focus of big tech competition?

I think there are three reasons.

First, office is a sufficiently high-frequency scenario.

Writing proposals, making PPTs, organizing meeting minutes, analyzing spreadsheets, looking up materials, and handling emails—these tasks happen every day. Compared with many low-frequency AI applications, office naturally has higher usage frequency.

Second, office has clear commercial value.

Whether AI can help an employee save two hours, whether it can reduce repetitive labor in a department, whether it can shorten a project's delivery cycle—enterprises can calculate these.

Third, office has commercialization space.

Individual users can pay for efficiency, and enterprises can pay for organizational efficiency. Especially when AI begins to enter enterprise knowledge bases, business systems, and internal processes, it is no longer just a software subscription, but has the opportunity to become part of enterprise digital infrastructure.

Node Finance believes that the value of AI office is migrating from a personal efficiency tool to an enterprise operation tool. Whoever can first connect the work chain among individuals, teams, and organizations will have the opportunity to build the competitive barrier of the next-generation office platform.

In the past, when we used AI, most of the time we asked it questions and then completed the follow-up work ourselves.

Now, the situation is changing.

We want AI to break down tasks itself, call tools, process files, analyze data, and finally deliver results directly.

From answering questions to completing tasks—this change looks small, but actually means a change in the entire AI application business model.

Six big tech companies, six different competition routes

If we look at the domestic AI office market together, Tencent, Alibaba, ByteDance, Baidu, Huawei, and Kingsoft Office are all doing AI office, but their resource endowments and entry points are not the same.

Tencent: from communication entry to enterprise-level Agent

For Tencent's AI office, the most noteworthy thing is still the combination of WorkBuddy and WeCom.

On October 14, Tencent announced that WeCom and WorkBuddy had achieved two-way deep integration. WorkBuddy can enter the WeCom address book and, within the scope of enterprise authorization, call work context such as messages, documents, meetings, and cloud drives.

This matter is very important.

Because much of an enterprise's work originally happens in communication. Requirements are raised in groups, tasks are assigned in groups, and results also need to be fed back to groups.

If AI can understand these communication contents and further complete file processing, information organization, and task execution, it has the opportunity to turn the communication entry into a work entry.

On October 30, Tencent brought four upgrades to WorkBuddy Enterprise Edition: open platform, enterprise-level security, observability center, and hosted agent WMA.

Tencent disclosed that since WorkBuddy launched in March, it has cumulatively completed more than 9 billion user task requests, covering more than 50 industries and 50% of China's Fortune 500 companies.

It should be emphasized that 9 billion is cumulative task requests, not the number of users.

Tencent's advantage is the ecosystem connectivity formed by WeChat, WeCom, and cloud services.

But it also has problems to solve: there is still a distance between chat scenarios and real business processes. AI being able to understand a conversation does not mean it can accurately complete complex tasks such as approvals, finance, and sales.

What Tencent needs to prove next is whether it can truly transform information in communication into organizational productivity.

Alibaba: from DingTalk collaboration to team Agent

On November 9, Alibaba Qianwen Office 2.0 was officially released, a product move in the AI office track that is very worth watching.

The keyword of this upgrade is Team Agent.

The new version supports collaboration among multiple people and Agents. Users can bring Agents into work groups, and it also adds capabilities such as human-machine co-creation, material space, automatic tasks, and three-terminal interconnection.

This means Alibaba hopes AI will no longer be just one person's assistant, but become a collaborator on the team.

Where is Alibaba's advantage?

DingTalk.

After years of development, DingTalk has accumulated infrastructure such as enterprise organizational structure, approvals, meetings, documents, attendance, and collaboration. If AI can enter these scenarios, it has the opportunity to obtain more complete enterprise context.

For example, after a meeting ends, AI not only generates minutes, but also extracts tasks, associates owners, and tracks project progress.

This is the office intelligence enterprises truly need.

Previously, Alibaba disclosed that Qianwen Office had more than 30 million users in its first month online, with enterprise users accounting for more than half. But this number is the cumulative user caliber for the first month online and cannot be directly compared with QuestMobile's monthly active data.

The problem Alibaba needs to solve is how to further transform product capabilities into stable enterprise delivery capabilities.

Enterprises will not pay long-term just because they Experience a new feature. Whether complex processes can run through, whether permissions can be managed well, and whether tasks can be completed stably determine whether it can enter the core business of enterprises.

ByteDance: can model capability and organizational collaboration form synergy?

ByteDance's layout in the AI office field has actually formed a multi-level product system.

One end is Doubao Work, and the other end is Feishu and Doubao Work Partner.

In August 2026, Doubao Work was officially released, emphasizing starting from goals, automatically breaking down tasks, calling tools, processing files, and completing delivery.

Starting August 14, Feishu's original aily was renamed Doubao Work Partner, further emphasizing AI entering the enterprise work environment and collaborating with employees.

On September 15, Feishu 8.0 was released, focusing on strengthening Agent capabilities.

This combination has its own logic: Doubao provides models and multimodal capabilities, Feishu provides organization, knowledge, and collaboration scenarios, and Agent is responsible for connecting the two.

ByteDance's advantage is model iteration speed, product capability, and content generation capability.

Especially as demand for content production such as research reports, marketing plans, images, and videos increases, ByteDance's models and product ecosystem have room to play.

But ByteDance also has a question to answer: personal-side product experience is not the same thing as enterprise-level service.

What enterprises need is not just a smart AI, but also permission isolation, data security, long-term stable operation, and the ability to be compatible with existing business systems.

Whether ByteDance can transform strong personal productivity capability into enterprise organizational capability is worth continued observation.

Baidu: knowledge and professional tasks are its entry point

For Baidu's AI office layout, what is worth watching is Baidu Dazi and Kuku AI.

Baidu Dazi is oriented toward complex task processing, covering scenarios such as research analysis, document generation, and data processing.

From Baidu's official update records, on October 11, Baidu Dazi already supported online editing of Word, PPT, and Excel documents; at the end of October, it continued product iteration.

Baidu's advantage lies in its accumulation in search, knowledge, libraries, and cloud drives.

One of the essences of office is processing information. A large number of enterprise employees need to read materials, organize materials, and form reports every day. If AI can connect information collection, understanding, analysis, and output into one chain, it can create practical value.

But Baidu also faces a problem: there is still a distance from knowledge processing to organizational collaboration.

Writing a research report does not mean being able to drive a project; analyzing a financial report does not mean being able to enter the enterprise operation process.

What Baidu needs to prove in the future is whether it can further extend its content and knowledge advantages into enterprise business systems.

Kingsoft Office: the traditional office entry is being restructured by AI

Compared with other internet giants, Kingsoft Office has a very special advantage: WPS itself is an office entry.

Many people use WPS every day, and documents, spreadsheets, and PPTs have already formed fairly mature user habits.

The latest disclosed Q3 2026 report also provides some noteworthy data.

In the first three quarters, Kingsoft Office achieved operating revenue of 5.080 billion yuan, a year-on-year increase of 21.57%.

Among them, WPS personal business revenue was 1.050 billion yuan, a year-on-year increase of 16.78%; WPS 365 business revenue was 270 million yuan, a year-on-year increase of 33.83%.

As of September 30, WPS global monthly active devices reached 678 million.

On the product side, Kingsoft Office launched Lingxi, a new AI-native office product, and continued to advance WPS Comate enterprise-level AI capabilities.

This set of data shows that Kingsoft Office is no longer just telling an AI story; AI functions are combining with personal payment conversion and enterprise business growth.

However, Kingsoft Office also has its own challenges.

The advantage of traditional office software lies in document editing, but the way of working in the AI era may change.

In the future, users may no longer need to open Word, Excel, and PPT separately, but directly propose a work goal, and AI will automatically call different tools to complete the task.

How Kingsoft Office transforms from a traditional office software provider into an AI-native productivity platform is a question worth long-term observation.

Huawei: HarmonyOS terminals provide another possibility

On November 6, Huawei Xiaoyi standalone App was officially launched on the HarmonyOS PC App Store, and Xiaoyi Work official version was launched simultaneously.

What is worth watching in this upgrade is that AI has begun to further enter the system usage scenarios of HarmonyOS PCs.

According to the public product introduction, Xiaoyi Work supports assigning tasks such as data analysis, PPTs, and weekly reports through one sentence, and generating Word, PPT, and Excel results.

At the same time, it can access office tools such as DingTalk and Feishu through connectors, support task execution on cloud and local computers, and users can also remotely assign tasks through HarmonyOS phones.

Huawei's advantage is relatively special: it has an operating system, terminal devices, and software-hardware collaboration capabilities.

This means AI office does not necessarily have to rely only on standalone software; it may also become a foundational capability at the operating system level.

But Huawei also needs to face problems of application ecosystem and enterprise services. Terminal entry is only the first step; whether it can connect enough enterprise applications and stably complete complex tasks determines its commercial value.

Six companies, six different approaches. There is no completely identical starting line among them.

The real competition in 2027 is not about who is better at making PPTs

Putting these products together, I think there is a very clear trend:

AI office is moving from single-point feature competition to complete workflow competition.

There is an essential difference between the two.

Today, almost all leading AI products can write articles, make PPTs, and organize spreadsheets. These capabilities are certainly useful, but they are also increasingly easy to replicate.

What is truly difficult is whether AI can fully participate in a piece of work from beginning to end.

For example.

A company is preparing for a new product launch.

The traditional workflow goes like this: marketing collects materials, product organizes product information, sales provides customer feedback, finance calculates the budget, management meets to discuss, and finally marketing puts together the launch plan.

That involves multiple departments, multiple documents, and multiple systems.

If AI only helps the marketing department write a PPT, it solves just one link in the chain.

But if AI can—within the scope of authorization—read materials, summarize sales feedback, analyze budgets, generate plans, and then turn meeting decisions into tasks and track them continuously, then what it changes isn't just one person's efficiency. It changes how the entire organization collaborates.

That's the truly promising part of AI in the office.

Of course, several hurdles remain.

The first is accuracy. AI can produce a report that looks highly professional, but whether the data sources are reliable and the calculations are accurate still needs verification.

The second is security. Enterprise customer materials, financial data, and business plans all require clear permission management and audit mechanisms.

The third is system integration. Enterprises typically run ERP, CRM, financial software, and various other business systems internally. Whether AI can actually connect to these systems matters more than whether it can generate a beautiful PPT.

Finally, there's the input-output ratio. Enterprises don't buy AI services to chase trends—they want to see how much cost it saves and how much efficiency it improves.

Node Finance believes that the real watershed in AI office competition in 2027 may not be who builds a smarter Agent first, but who can make the Agent work reliably inside enterprises and continuously create value.

The endgame of AI in the office may be a change in how enterprises are organized

There's another issue worth watching: what AI in the office ultimately changes may not be just software, but the enterprise organization itself.

In the past, how many people an enterprise needed often depended on business scale and management span.

A project requires collaboration across roles—marketing, sales, operations, finance. A lot of time isn't spent creating value but on information transmission, material organization, repetitive communication, and process coordination.

Once AI enters these links, the division of roles, management methods, and talent structure of enterprises may all change.

For example, a marketing person used to need half a day to organize industry materials. Now they can have AI handle material collection, preliminary analysis, and report framing first, and put more energy into judgment and strategy.

A manager used to have to repeatedly ask about project progress. Now, through an enterprise-authorized AI system, they can stay on top of relevant information in real time.

A team used to need multiple people to pass files back and forth. Now an Agent can take on part of the information organization and process coordination work.

Of course, this doesn't mean AI will replace a large number of employees tomorrow.

The truly realistic change is that one employee, with AI's help, can complete part of the work that previously required collaboration among multiple people.

For business owners, the more important question in the future isn't whether to purchase an AI office software suite, but which processes are worth redesigning, which positions need to be redefined, and which data can truly be utilized by AI.

Node Finance believes that the biggest business opportunity in AI office work may not lie in software subscription fees themselves, but in whether it can help enterprises reorganize productivity.

This also means that in the future, competition among enterprises will include an important additional dimension beyond capital, talent, and products: the organization's ability to use AI.

Final Thoughts

Looking back at two decades of internet development: in the PC internet era, the entry points were portals and search; in the mobile internet era, the entry points were super apps.

In the AI era, the office entry point may become the next major competitive battleground.

Tencent has WeChat and WeCom, Alibaba has DingTalk, ByteDance has Feishu and Doubao, Baidu has its knowledge and content ecosystem, Kingsoft Office has WPS, and Huawei has HarmonyOS terminals.

They hold different cards and take different paths.

But one thing is the same: none of them want AI to remain merely in a chat window.

Node Finance believes that in 2027, the real decisive factor in AI office work won't be model parameters, the number of features announced at launch events, or short-term user scale. It will come down to three words: delivery capability.

In 2027, the great drama of AI in the office has only just begun.