Facing Direct Competition from Tencent's WorkBuddy

Author｜Yixiu

Editor｜Xiaobai

Illustration｜AI Generated

Produced by｜Qiangdiao Next

On July 27th, the official website of Qianwen Office was officially launched, with downloadable versions for Windows, macOS Beta, and HarmonyOS PC available simultaneously. The web version and the DingTalk integrated version are set to follow later.

This marks the first publicly released product by Chen Yusen since he assumed the role of CEO of DingTalk on June 11th, a process that took just 46 days.

According to the official website, Qianwen Office's personal version is offered in three tiers: free, 98 RMB, and 198 RMB, with ongoing monthly subscription options priced at 78 RMB and 158 RMB. The enterprise standard version is priced at 198 RMB per month per seat, with the flagship version requiring a customized quote. The capabilities listed on the official website encompass enterprise IM collaboration, Office content generation and editing, multimodal processing, webpage generation, professional data source access, and a skill marketplace, among others.

The early launch, before all products are fully prepared, underscores the significant pressure Tencent has exerted on Alibaba. Data from Analysys reveals that in June of this year, the traffic of intelligent agents under Tencent was approximately 3.5 times that of Alibaba's.

Clearly, 46 days is insufficient to develop a fully mature office agent from the ground up. During this period, Chen Yusen's primary focus was on identifying Alibaba's main office agent product, integrating teams, and formulating a commercialization strategy.

1 · 46 Days: Securing the Lead

In early 2026, OpenClaw spurred the growth of office agents, leading to the emergence of three products within Alibaba: QoderWork, Wukong, and MuleRun.

Each product had its unique strengths. QoderWork, an extension of the programming tool Qoder, excelled in handling local files, invoking desktop tools, and delivering office outputs. Wukong, developed by DingTalk, was more aligned with enterprise organization and collaboration. MuleRun focused on 24/7 cloud operation, scheduled tasks, and skill reuse.

However, the overlapping positioning of these three products and their affiliation with different teams posed a challenge. Continuing in parallel would fragment user entry points, R&D resources, and computing power investments.

A week after taking office, Chen Yusen issued his first company-wide memo, announcing the integration of the Wukong and MuleRun teams, while consolidating direct sales, telesales, presales, delivery, and customer success teams into a Customer Development Department.

In early July, QoderWork was also integrated. The final Qianwen Office was built on QoderWork, incorporating features from Wukong and MuleRun.

This decision was based on practical considerations.

Analysys data indicates that in June 2026, the 17 domestic desktop AI-native office intelligent agents collectively recorded over 60 million visits. Tencent's WorkBuddy accounted for 20.97 million, ByteDance's TRAE IDE domestic version for 12.79 million, and QoderWork for 7.88 million. QoderWork had the strongest user base within Alibaba.

Notably, Chen Yusen did not directly promote MuleRun, which he had previously incubated. Instead, he chose QoderWork as the front-end, breaking MuleRun down into cloud operation and skill capabilities.

This approach reveals Chen Yusen's product strategy after assuming office: cease redundant competition, identify the main product based on user feedback, and then allocate teams and resources accordingly.

Qianwen Office thus established a clear division of labor. QoderWork handled desktop interaction and local task execution, MuleRun provided continuous cloud operation capabilities, Wukong connected enterprise scenarios, and DingTalk supported organization, permissions, data, and customer systems.

The standalone client is merely a product manifestation under this structure and does not warrant excessive strategic analysis. What truly matters is that Alibaba has finally prioritized its scattered office agents.

2 · The First Product: Directly Assessing Paid Adoption

Qianwen Office announced its full pricing structure even during its Beta phase.

The personal premium version costs 158 RMB for a continuous monthly subscription, while the enterprise standard version is priced at 198 RMB per month per seat. This pricing places Qianwen Office in the realm of professional productivity software. Rather than relying on free adoption for growth, Alibaba is simultaneously testing user demand and willingness to pay.

This strategy also means Qianwen Office cannot compete solely on model capabilities.

Generating PPTs, writing documents, analyzing spreadsheets, and creating webpages have become standard features of office agents. As model capabilities converge, product differentiation shifts to task execution: the ability to read local and enterprise data, reliably invoke tools, control operational permissions, and directly integrate results into collaboration and approval workflows.

Alibaba already possesses these resources.

Qianwen provides the model, QoderWork offers the desktop entry point, MuleRun supplies the cloud execution environment, and DingTalk brings organizational relationships, enterprise workflows, and a customer base. Chen Yusen's task is to transform these capabilities from a mere product list into a seamless workflow for users.

If users still need to repeatedly authorize, upload files, and copy results across different products, Qianwen Office would merely be a brand consolidation. Only when tasks can run continuously across local computers, cloud execution environments, and enterprise systems does this integration truly succeed.

Charging fees will compel Qianwen Office to address these issues promptly.

Individual users may tolerate occasional errors, but enterprises purchasing hundreds of seats prioritize task completion rates, data security, permission controls, and result traceability. Selling at 198 RMB per month per seat cannot rely solely on a set of AI features; it must include reliable enterprise delivery.

3 · Launch Is Just the First Step

Tencent has already established a first-mover advantage in the office agent market. Analysys data shows that WorkBuddy recorded 20.97 million visits in June this year, approximately 2.7 times that of QoderWork. On July 20th, the QClaw product center's related businesses and some teams were transferred to Cloud Product Division VI, where WorkBuddy resides, as Tencent also consolidates its resources.

Qianwen Office's launch at this time is primarily a catch-up effort.

Chen Yusen has already addressed three initial challenges: which product to prioritize, how to consolidate teams, and how to price the product. The next step is to verify how well the three products integrate and whether DingTalk's enterprise capabilities can truly connect with Qianwen Office.

Success cannot be measured by traffic alone. Metrics such as active task users, task completion rates, enterprise seat activation rates, and subscription renewal rates better indicate whether the product succeeds.

Currently, Qianwen Office is still in limited testing, with the web version and the DingTalk integrated version yet to launch. The user rights, data, and teams of the original products also require further integration. Chen Yusen has merely submitted his first-stage answer sheet.

However, his initial move is clear: select the product with the strongest existing user base, eliminate internal indecision, concentrate resources, and verify demand through charging as early as possible.

The 46-day timeline reflects decision-making speed. Whether Qianwen Office can succeed depends on whether Chen Yusen can compress Alibaba's existing model, agent, and enterprise service capabilities into a product that delivers reliably and sustains subscriptions.

- END -