Alibaba is preparing to launch Qianwen Office, an AI-powered productivity suite that will consolidate three separate agent products into a single platform, intensifying a three-way battle with Tencent and ByteDance for China's $80.4 billion AI office market.
Alibaba Group is preparing to launch Qianwen Office, an AI-powered productivity suite that consolidates three separate agent products into one platform, intensifying a three-way battle with Tencent Holdings and ByteDance for China's $80.4 billion AI office market.
"The integration brings together QoderWork, Wukong and MuleRun under a single brand, pooling resources that were previously spread across competing internal teams," a person familiar with the matter said, asking not to be identified discussing private plans.
The product, already in internal testing, is positioned independently from Alibaba's Tongyi Qianwen large language model and focuses on deep office scenarios including document creation, spreadsheet analysis and email automation. Alibaba has begun recruiting solution architects, business development experts and AI product managers nationwide, signaling a commercial launch is imminent.
The move comes as China's AI agent market expands at 123 percent annually, with iiMedia Research projecting it will reach 696.8 billion yuan by 2030. Alibaba's combined desktop traffic from its existing agent products stood at about 9.19 million monthly visits in June, trailing Tencent's 32.62 million and ByteDance's 14.41 million, according to Analysys.
Consolidation replaces internal competition
Alibaba's decision to merge its three agent products follows a pattern playing out across China's technology giants. Tencent has already consolidated its QClaw product center into the same division as WorkBuddy, its leading desktop agent that generated 20.97 million monthly visits in June — more than the next two competitors combined. ByteDance is pursuing a similar strategy, with internal discussions underway to deeply integrate its Doubao AI assistant with the Feishu enterprise collaboration platform, according to media reports.
The consolidation reflects a recognition that the window for capturing market share is narrowing. In the first half of 2025, China added 2.86 million new one-person companies, bringing the total to 16 million, according to iiMedia Research — a demographic that represents a natural customer base for AI office tools that replace administrative staff.
Model strategy diverges among the three players
Each company is taking a different approach to the underlying AI technology. Tencent's WorkBuddy operates as a model aggregation platform, integrating DeepSeek, Hunyuan, GLM and Kimi models that users access through a points-based system. On July 6, Tencent launched its Hunyuan Hy3 model and connected it to WorkBuddy; within weeks, more than 60 percent of users who selected their own model chose Hy3, according to Analysys.
Alibaba's strength lies in its model capabilities. Tongyi Qianwen is widely considered the most capable of the three companies' in-house large language models, giving Qianwen Office a potential advantage in task quality for complex document and data analysis work. However, the company's C-end distribution lags behind Tencent's WeChat ecosystem and ByteDance's Doubao user base.
ByteDance's TRAE IDE ranks second among desktop office agents with strong performance in programming tasks, but the product's positioning is ambiguous — it competes more directly with coding tools like Cursor than with general office agents. The company's bet on integrating Doubao with Feishu represents a different path, embedding AI capabilities directly into the workflow rather than requiring a separate client download.
Investment implications
Alibaba shares trade at about 10 times forward earnings, a discount to Tencent's 15 times, reflecting investor skepticism about the company's ability to monetize its AI investments. A successful Qianwen Office launch could narrow that gap by demonstrating a clear path to enterprise AI revenue. Tencent's WorkBuddy has already shown that desktop agent traffic can convert to paying users through its points-and-membership model, though neither company has disclosed specific revenue figures for its AI office products.
The biggest risk for all three players is that the market consolidates around a single winner, leaving the others with stranded investments in product development and model training. With the total addressable market projected to reach nearly 700 billion yuan by 2030, the stakes are high enough that none of the three can afford to sit out — but the window for establishing a durable competitive position may close within the next two quarters.
This article is for informational purposes only and does not constitute investment advice.
