Universal office AI agents represent one of the core tracks for the commercial implementation of large language models, currently at an inflection point transitioning from product validation to widespread adoption.

Alibaba-W (09988) has taken a leading position with its flagship product, QoderWork, being among the first to achieve productization and public testing of a desktop-level agent. The company has positioned it at the group level as the core interface bridging large models with the digital world.

Currently, Alibaba is advancing the deep integration of its three major To-B AI agents: QoderWork, Wukong, and MuleRun, thereby strengthening its strategic footprint in the universal office agent sector. Monitoring the progress of AI agents in the general office domain and Alibaba's advancements in this field is recommended.

Analysis of QoderWork

QoderWork is a foundational, application-layer product from Alibaba's ATH business group, launched as a desktop AI agent for non-technical staff. It stands out as the fastest among major domestic tech firms to achieve technological translation and comprehensive public testing for a general office agent.

Alibaba has defined QoderWork as the pivotal interface connecting large models to the digital world, operating under the ATH business group directly overseen by CEO Wu Yongming. In June 2026, Wu disclosed that its daily active users, weekly active users, and token usage ranked first among all the group's AI tools.

In July 2026, Alibaba initiated a deep merger of its three To-B agents, using QoderWork as the foundational platform and integrating Wukong and MuleRun under the unified leadership of Chen Yusen.

Market Outlook for Universal Office Agents

The market holds potential on a scale of tens of billions, with significant elasticity in both user penetration and willingness to pay. Data from leading firms indicates strong growth in this segment, with use cases increasingly tied to core business processes.

Two top global vendors have recently adjusted their product architectures, significantly elevating the strategic priority of office agents, which validates the sector's potential. The Chinese market remains in the very early stages of the adoption curve.

Benchmarking against the current penetration rates of general AI tools, the estimated addressable market for a mature state in 3-5 years is approximately 39 billion yuan annually. Broad user penetration and corporate payment adoption are the primary sources of market elasticity, with optimistic projections suggesting a potential market size reaching hundreds of billions.

Competitive Strengths of QoderWork

QoderWork has established a differentiated competitive advantage across three dimensions: its underlying model, tool invocation capabilities, and ecosystem strength.

At the model level, QoderWork follows a vertically integrated path using self-developed models. Its Qwen3.7-Max model ranks among the top domestic models on benchmarks like LMArena and Artificial Analysis, demonstrating exceptional performance in reducing hallucinations, following instructions, and handling enterprise process operations.

Regarding tools, it employs a two-tier architecture of skills and expert suites, covering nearly 20 professional domains. This approach combines general skills with industry-specific knowledge to form actionable knowledge skills.

Ecosystem-wise, leveraging Alibaba's four major platforms (DingTalk, Amap, Alibaba Cloud, 1688.com), augmented by 26 connectors and 9 IM channels, it has built an integrated system covering mainstream office tools.

Key Risk Factors

Potential risks include the performance of universal office agent products falling short of expectations; slower-than-anticipated growth in broad user penetration; intensified market competition putting pressure on pricing levels; and self-developed model iterations lagging, which could constrain the product's capability ceiling.