For the first time, Tencent WorkBuddy has a dedicated PC, and ASUS is the first to take the plunge.

ASUS recently launched a WorkBuddy exclusive PC, becoming the first PC manufacturer to launch a WorkBuddy exclusive computer.

Compared to users downloading WorkBuddy themselves, this product further integrates the Agent with local models, Skills, and PC hardware capabilities.

This isn't the first time ASUS has incorporated an office agent into its AI PCs. In June of this year, ASUS commercial PCs partnered with Baidu AI Cloud to launch the "ASUS × Baidu DuMate AI Solution," which also emphasizes the combination of edge computing power and agent capabilities.

Meanwhile, hardware manufacturers such as Honor, Lenovo, and Xiaomi are also extending their Agent capabilities to PCs. Honor pre-installed YOYO Claw into the MagicBook, Lenovo continues to upgrade Tianxi Claw, and Xiaomi brought Super Xiao Ai 2.0 "Expert Mode" to PCs.

As edge-cloud collaboration matures, providing an agent within a computer that can directly process files, call tools, and even execute tasks is becoming a new selling point for AI PCs.

Whether the PC's built-in agent will ultimately come from a third party or be developed by the hardware manufacturers themselves remains undecided. What truly determines this path may be who first develops a sufficiently high-frequency, sufficiently universal "national-level agent."

When office agents become a selling point

The core differences between this ASUS Workbuddy exclusive version and the version where users download WorkBuddy themselves lie in two main points:

First, there's a local skill adapted for the Intel platform; second, there's deeper collaboration between the local model and the cloud model.

Currently, Intel has already launched 15 local skills in the ASUS-exclusive WorkBuddy. On the ASUS Dawn high-end series equipped with 3rd generation Intel Core Ultra processors, even more local models can be utilized. Third-party developers have already locally adapted and optimized 26 WorkBuddy high-frequency experts based on the 35B-A3B local model.

In practical use, some local file processing can be done directly on the computer; when encountering complex inference, long-chain tasks, or when real-time information is needed, the cloud model is then invoked. WorkBuddy will allocate resources between the two based on the task requirements.

Ma Yingkui, head of Tencent Cloud's intelligent manufacturing industry solution in East China, told Wall Street Insights and All-Weather Technology that Tencent WorkBuddy allows users to flexibly choose from a variety of models; users can also choose local or cloud processing, and switch to the cloud if they are not satisfied with the local processing results.

Ma Yingkui further pointed out that users no longer need to determine "what kind of model to run on this PC to maximize its performance." Running some tasks locally can also reduce cloud resource usage, and data that users don't want to upload, such as contracts and financial documents, can be processed on the computer.

Currently, ASUS Workbuddy Exclusive Edition primarily targets two groups: one is small and medium-sized enterprises and entrepreneurs who do not have a professional AI team but want to quickly use the Agent; the other is "super individuals" who use AI more frequently, such as content creators who can complete copywriting, video generation, editing, and multi-platform publishing all by themselves.

For PC manufacturers like ASUS, the office agent also provides a new selling point for AI PCs.

WorkBuddy is not the first office agent that ASUS has partnered with this year.

In June of this year, ASUS commercial PCs and Baidu AI Cloud launched the "ASUS × Baidu Partner DuMate AI Solution". This solution is also based on the edge computing power of ASUS commercial AI PCs, combining it with the Agent capabilities of Baidu Partner, and is mainly aimed at enterprise office scenarios.

This also illustrates that PC manufacturers shouldn't prematurely bet on a single company before the performance of office agents has fully emerged. Rather than simply competing on chip specifications, those who can transform the computer's existing local computing power into features that users will actually use, or even create new reasons to upgrade, are more likely to gain deeper adaptation resources from PC manufacturers.

However, for Tencent, ASUS is more like a demonstration project.

In early September, the WorkBuddy open platform was launched, initially introducing more than 100 ecosystem partners and opening up its capabilities to smart hardware, industry applications and developers.

At the time, Liu Yi, Vice President of Tencent Cloud and head of Tencent CodeBuddy & WorkBuddy, set the goal of making WorkBuddy an "operating system for the Agent era," rather than just a more powerful tool. Its hardware partnerships currently cover more than ten categories and over 30 brands, including PCs and tablets, smart glasses, and headphones.

To achieve this goal, simply making one ASUS WorkBuddy exclusive computer is clearly not enough.

Whether Tencent will collaborate with more PC manufacturers to launch exclusive Workbuddy PCs is attracting more attention.

II. Self-built vs. Third-party

For the long-dormant PC market, putting a real agent into a computer could indeed become a new product narrative.

ASUS has chosen to partner with third-party agents such as WorkBuddy and Baidu. In addition, some PC manufacturers with hardware, operating system, and AI capabilities tend to extend their agents directly to the computers.

This year, Honor pre-installed its self-developed YOYO Claw on the MagicBook Pro series 2026, emphasizing automatic deployment and "out-of-the-box" functionality. It can autonomously plan, schedule, and execute tasks, and integrate with various skills such as office work, learning, and social media. The subsequently released MagicBook 14 and MagicBook 16 2026 also listed the YOYO AI agent as a major selling point.

Lenovo got there earlier.

In March of this year, Lenovo released Tianxi AI Claw, emphasizing "zero-cost deployment and zero-threshold use"; in May, it further released Tianxi AI 4.0, upgrading the core of the intelligent agent to Tianxi Claw and continuing to strengthen the "device-edge-cloud" collaboration.

According to Lenovo's plan, complex tasks can be allocated among different computing powers, while also supporting the local execution of large models.

Xiaomi is also extending its Agent capabilities to PCs.

With the release of HyperOS 4, Super AI 2.0 has added an "Expert Mode" that can complete tasks such as in-depth research, PPT creation, and web page creation. It has been adapted to PCs such as Xiaomi Book Pro 14 and REDMI Book Pro 16 2026 in the first batch.

The paths taken by the different companies are not exactly the same, but the underlying considerations are similar.

For PC manufacturers, it's easier to establish a product selling point by directly handing over an agent that can access files, software, and models to users.

The maturity of end-to-end cloud integration has lowered the barrier to entry for this endeavor.

Agents don't need to cram all their capabilities into local storage. Tasks involving privacy and simple file processing can be completed on the client side first, and the cloud can be invoked when larger models and real-time information are needed.

This means that if an agent that can be used immediately upon booting up and can actually perform some tasks for the user can be provided in a PC at a relatively controllable hardware and cloud cost, it may indeed become a new selling point for PC manufacturers in the next stage.

Whether this agent comes from a third party, is created by the hardware manufacturer itself, or is a combination of both, there is currently no answer.

Ultimately, which approach will become the choice of more manufacturers will likely depend on which agent can truly demonstrate stable user habits.

Risk warning and disclaimerInvesting involves risk; please exercise caution. This article does not constitute personal investment advice and does not take into account the specific investment objectives, financial situation, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article are suitable for their specific circumstances. Any investment decisions made based on this information are at your own risk.