Dialogue with the Tencent WorkBuddy leader: An ambitious product aiming to rival Cowork
```
Author | Huang Yu
From the sudden emergence of OpenClaw at the beginning of the year, to tech enthusiasts eagerly sharing their “shrimp farming” experiences on social media, and then ordinary people lining up at Tencent headquarters for free OpenClaw installations.
In the spring of 2026, a “lobster storm” triggered by desktop Agents is quietly reshaping the working paradigms of every office worker.
In this intense tide of efficiency revolution, Tencent’s WorkBuddy officially launched on March 9. User visits soared, causing WorkBuddy to face the “growing pains” of overwhelming demand by the second day, being forced to urgently expand tenfold to accommodate market enthusiasm.
This scenario of “crashing upon launch” reveals the high enthusiasm non-developers have for “desktop Agent-type products.”
As the product lead for Tencent WorkBuddy, as well as the head of Tencent Cloud Developer Products, Wang Shengjie witnessed the entire journey of WorkBuddy from incubation to explosive popularity.
On March 13, Wang Shengjie revealed in exchanges with media such as Wall Street Insights that WorkBuddy’s market performance far exceeded team expectations after its launch, with requests surpassing CodeBuddy many times over, and computing power was far insufficient, reaching warning thresholds. “We urgently expanded capacity and optimized the architecture to make login logic more stable.”
WorkBuddy was developed by Tencent Cloud’s CodeBuddy team. CodeBuddy is an AI programming tool released by Tencent last year, and WorkBuddy is built on the same Agent architecture as CodeBuddy—a desktop intelligent agent platform with a foundation, an ecosystem, and expandability.
The birth of WorkBuddy was not a rushed trend-following act, but a “blitzkrieg” that had long been planned.
As early as the second half of 2025, Wang Shengjie’s team had already built the underlying architecture and open platform for CodeBuddy to support AI autonomous task execution, and had released the SDK. The strategy of solidifying execution infrastructure first and then opening the ecosystem mirrors the core logic behind Anthropic’s Claude Cowork.
Wang Shengjie recalled, “After New Year’s Day this year, Claude Cowork came out, so I took my idea to the boss. Based on our own platform, we quickly iterated to produce the prototype of WorkBuddy. The boss agreed immediately and thought it was great, so we got started.”
Wang Shengjie specifically mentioned that before OpenClaw became popular, the strongest product in this sector was actually Claude Cowork, but it didn’t catch on because it was overseas and not open source.
After just two nights of all-nighters, version 0.01 of WorkBuddy went online, though at that time it was mainly for use within Tencent. “I remember it was January 17 when we started, those two days were Saturday and Sunday, and a few of us pulled all-nighters for two days.”
The outside world first learned that Tencent was developing a desktop Agent tool similar to OpenClaw on February 6.
That day, the official account of “Tencent Cloud Code Assistant CodeBuddy” officially announced the start of WorkBuddy’s internal testing.
With OpenClaw’s explosive popularity after the Spring Festival, the CodeBuddy team accelerated WorkBuddy’s official launch process.
Wang Shengjie explained that the key to OpenClaw’s sudden popularity was its combo of open-source capabilities and automation for office scenarios. WorkBuddy and OpenClaw share very similar missions, but their approaches differ.
In simple terms, WorkBuddy takes a self-developed, product-oriented, safe, controllable, and ecosystem-integrated path, while OpenClaw takes an open-source, highly flexible, community-driven route.
Wang Shengjie said: “When we incubated WorkBuddy internally, we focused more on experience and security, helping users solve their problems within a controlled scope.”
He stated that WorkBuddy is 100% self-developed and has not used a single line of OpenClaw source code. “Because we already have the entire CodeBuddy infrastructure—AI-driven, autonomous task completion framework, so there’s no need to borrow. We started from user scenarios to build a product aligned with the OpenClaw model, not just a shell.”
As the focus of the AI industry competition shifts from “model parameter contests” to “real-world application ability contests,” Agent products are in the spotlight. Today, OpenClaw seems to have become synonymous with such products, and any similar product launch is liable to be seen as following OpenClaw’s lead.
Meanwhile, Wang Shengjie clarified that lobster is a concept with different implementation strategies. WorkBuddy is not a purely managed automation product; in terms of automation degree, it can be considered in the middle position.
The WorkBuddy team’s assessment is: most users currently truly need search-plus-content-recombination scenarios, like daily AI news analysis, automatically posting Twitter content in Xiaohongshu (Little Red Book) style.
“That’s already a very advanced scenario, and also a safe one.” In Wang Shengjie’s view, there’s no need to let AI autonomously post or learn without user authorization—it’s a scary prospect.
“Overseas, some products attempt higher autonomy, even fully managed directions. When everyone truly understands which scenarios are valuable, WorkBuddy will then lay out fully automated capabilities.”
OpenClaw thoroughly popularized the lobster concept—not just a specific product name anymore, but a productivity concept symbolizing “autonomy, remote work.” But competition after the dividends fade is often accompanied by questions about safety and boundaries.
In this unpredictable technological life-or-death battle, whoever truly understands users’ pain points will be able to breed the strongest “lobster” in the jungle built on computing power.
Risk Warning & DisclaimerThe market has risks, investment needs caution. This article does not constitute personal investment advice, nor has it considered the particular investment goals, financial situation, or needs of individual users. Users should consider whether the opinions, views, or conclusions herein fit their specific circumstances. Investment based on this, responsibility is at your own risk. ```