Summary of Key Points
The field of AI-powered office solutions has become a battleground for the giants (Tencent, Alibaba, ByteDance, etc.), but previous analyses have already covered fundamental issues such as the collective entry of these companies, the evolution of agents from mere responders to task completers, and the importance of external variables affecting model performance. This article focuses on three deeper questions:
1. Why does the Chinese internet experience periodic “all-out battles” every few years?
2. When intelligence becomes a readily available “public tool,” where do the new scarce resources lie?
3. When everyone’s capabilities are roughly the same, what constitutes a truly irreplaceable advantage?
The article also predicts the competitive paths of the giants by 2026 and identifies five key indicators that will determine future winners and losers.
Why Do the Internet Battles Occur Periodically? — It’s About the Migration of Scarcity
Chinese internet companies engage in major battles every few years (in areas like food delivery, car-hailing, community group buying, AI-powered office solutions, etc.). The essence of these battles is a cycle of:
- Capital chasing scarce resources → Investing to increase supply → The old scarcity becoming the norm → New scarcity emerging:
- Food Delivery: Initially, there was a shortage of riders, so capital subsidized them to attract more; once the number of riders increased, the competition shifted to the national distribution network (which is difficult to replicate).
- Shared Bicycles: Initially, there were few bikes, so capital flooded the market; once the bikes were plentiful, the focus shifted to maintenance and efficiency (scheduling, loss control).
- AI-powered Office: Initially, there were a lack of large models; now that models are available through APIs (which anyone can purchase), the new scarcity has shifted to the backend systems that enable intelligent assistants to function reliably (e.g., context understanding, permission management).
In simple terms, capital is like a hunter chasing rabbits. Once one rabbit is caught, it moves on to the next one—once the old scarcity is satisfied, it seeks the next.
Intelligence Has Become a Public Resource, but Where Are the New Scarce Resources?
Previously, large models were exclusive assets; now, intelligence has become a purchasable resource. This has led to three changes:
1. Application capabilities may become obsolete: What’s impressive today could become standard tomorrow due to model upgrades (just like how phone cameras have evolved from 5 megapixels to 100 megapixels).
2. Old office tools are becoming shareable: Documents from platforms like Lianshu and DingTalk are no longer exclusive to their users; third-party AI products can access them through APIs (e.g., Qianwen can access Lianshu documents), but the control remains with the original platform (e.g., who can access them and to what extent).
3. Competition shifts to the underlying infrastructure: With everyone having models, the focus is on the backend systems that enable intelligent assistants to work effectively (e.g., whether they can remember your previous work progress, execute tasks across systems, and manage data according to company policies).
When Capabilities Are Easily Replicable, What Constitutes an Irreplaceable Advantage?
By “irreplaceable advantage,” we mean a capability that keeps users loyal even when 90% of the functions are copied by competitors. Based on this criteria:
- Models, functions, and traffic are not enough: Models can be purchased, functions may become obsolete, and switching between them is inexpensive (you can use both DouBao and Kimi).
- New value in old office tools: It’s not about how much data you have, but about who controls how that data is used (e.g., Microsoft’s WorkIQ provides work context to third parties while maintaining control over user identities and permissions).
- More substantial advantages include:
- Work context: Intelligent assistants know the progress of your projects (e.g., which materials have been excluded, which numbers need verification), not just having read your documents.
- Continuous task assignment: Users prefer to use the same assistant for tasks (e.g., they’ll use DouBao to write weekly reports instead of switching tools each time).
What Will the Giants Be Testing by 2026? — Three Different Competitive Paths
By 2026, the competitive strategies of the giants will have diverged:
1. Agentification of old office tools: Alibaba (DingTalk + Qianwen) will transform existing approval processes into tasks that intelligent assistants can execute; Kingsoft (WPS Lingxi) will create project spaces to keep documents and conversations related to projects.
2. Superfronts for deeper integration into work: Tencent (WorkBuddy) will enhance memory functions and open its platform for other apps to use its intelligent assistants; ByteDance (DouBao + Lianshu) will merge personal AI tools with corporate collaboration tools to guide users from personal to professional work.
3. AI-driven environment construction: Baidu (ZhaZi) will develop AI solutions for specialized fields like finance and legal services; Kimi Work will create project folders and permission systems from the ground up.
What to Look For in the Future: Five Indicators of Success
The ultimate winner in the AI-powered office battle will be determined by these five factors:
1. Migration cost: How difficult is it to switch to a new AI system without losing project progress or historical data?
2. Third-party integration: How often can third-party AI tools interact with existing office tools (e.g., can they not only read but also modify documents or initiate approvals)?
3. Route control: Who decides which models/tools are used for tasks (e.g., the AI automatically selects the best model and data source for a report).
4. Governance: Who controls the permissions of AI systems (e.g., whether companies or AI products have control over sensitive data).
5. Payment model: Will companies switch from buying individual accounts to paying for specific task results?
Any change in these indicators could alter the outcome of the battle—current giants may not win, and new players may still have a chance.
The core message of this article is that the competition in AI-powered office solutions is not about having the most features, but about controlling the scarce resources that truly make a difference: work context, continuous task assignment, and control over user interactions. The future winner may not be the one with the most traditional office tools, but the one that can establish a dominant position in the new technological landscape.