On August 25th, ByteDance introduced its standalone AI office solution, Doubao Work, marking a strategic shift as China’s leading internet companies redirect their focus toward enterprise AI after investing billions of yuan chasing consumer users.
Doubao Work is an agent-powered tool deeply integrated with ByteDance’s workplace collaboration platform, Lark. It can autonomously analyze tasks, activate various tools, and manage complex workflows. With proper user authorization, it can even control computers across different software programs. Additionally, the teams responsible for ByteDance’s AI coding platform TRAE and the AI agent development platform Coze will be integrated fully into the Doubao ecosystem.
Major Chinese tech firms are realigning their AI product offerings around workplace solutions, aiming for clearer demand and more sustainable revenue streams after costly consumer-focused campaigns that resulted in only short-term traffic boosts. However, prospects remain uncertain, as Chinese enterprises are less accustomed to paying for SaaS products, and enterprise AI must demonstrate its ability to reduce costs and enhance productivity.
This paradigm shift is already reshaping product lines industry-wide. Tencent has consolidated its remote computer operation AI agent, QClaw, into the same structure as its workplace agent, WorkBuddy. Alibaba has combined its AI tools QoderWork, Wukong, and MuleRun into the unified platform Qwen Office, centralizing various AI workplace solutions.
From Consumer Engagement to Corporate AI Solutions
During the Chinese New Year holiday in February, these internet giants were still heavily investing in consumer AI initiatives. Tencent’s Yuanbao, a versatile AI assistant, launched a CNY1 billion (roughly USD148.8 million) cash giveaway campaign. ByteDance’s chatbot Doubao partnered exclusively with China Central Television’s Spring Festival Gala. Similarly, Alibaba’s Qwen AI assistant rolled out a CNY3 billion “Spring Festival Treat Campaign,” offering free food, shopping, and entertainment to users.
This large-scale spending resulted in peak user engagement. Doubao’s daily active users reached 145 million, Qwen attracted 73.5 million, and Yuanbao had 40.5 million. Yet, after the promotional campaigns ended, user activity numbers plummeted. Their rankings in major app stores fell rapidly, and user retention and conversion rates fell far below expectations.
The core challenge lies in the fundamentally changed economic logic of AI. In the era of mobile internet, spending big to attract traffic was profitable because marginal costs were near zero, and revenue could be generated through advertising. For large AI models, however, every user interaction requires significant computing power, increasing operational costs without the same network effects that drove growth in previous digital eras.
The financial impact of this AI-driven push is already evident. Tencent’s operating profit under non-IFRS standards hit CNY75.6 billion (about USD11.3 billion) in the first quarter, showing a 9% increase from the previous year, according to the company’s financial statements.
Excluding costs related to five new AI products—Hy, Yuanbao, CodeBuddy, WorkBuddy, and QClaw—non-IFRS operating profit was approximately CNY84.4 billion, up 17% year-over-year. The difference of CNY8.8 billion highlights the expenses linked to these new AI initiatives.
These costs are pushing tech giants toward solutions where customers are more willing to pay for services.
Office AI’s Commercialization Hurdles
However, workplace AI still faces significant challenges in proving its profitability. Zhang Fa’en, lead scientist at enterprise AI solutions provider AInnovation Technology Group, previously explained that the mature SaaS ecosystem in the U.S. largely involves agents replacing “human operators” of software. In China, however, the culture of paying for SaaS remains underdeveloped.
Procurement decisions in enterprises are primarily aimed at cutting expenses and increasing efficiency, not technology experimentation. To be truly valuable, AI solutions need to be deeply integrated into specific business workflows, demanding high levels of industry knowledge, security, and customization from the providers.
Therefore, enterprise AI office products must demonstrate tangible benefits before gaining widespread acceptance. Only then can they transition from experimental tools to essential business solutions.