Hello! I'm your financial and business news analysis assistant. This press release is packed with a lot of information and is logically very well-structured. It's not just about who merged with whom; it explores a deeper industry issue: the paradox of the cost and value of AI Agents in terms of their context.
To make it easier for you to understand, I'll translate this complex B2B business analysis into plain language and break it down into five key aspects.
---
📝 Summary of Key Points: A Movement Towards Centralized AI Functions
In one sentence:
In the summer of 2026, the six major global tech giants—Anthropic, OpenAI, ByteDance, Alibaba, Tencent, and Baidu—almost simultaneously took action to consolidate their scattered AI functions into a single interface.
Why?
On the surface, it seems to be about making the user experience smoother by reducing the number of buttons. But the real reason is that AI Agents rely on context to function effectively, and context is extremely fragile when it's fragmented. If AI knows about your files in one app but not in another, it has to relearn everything, which is inefficient and costly in terms of computing power. Therefore, the giants need to unify the interfaces so that AI can have access to all your permissions and data within a single, integrated system, thus saving money and improving usability.
---
🔍 In-Depth Analysis: Five Key Dimensions
1. **Action Analysis:表面上 Merging, but Actually Each is Doing Their Own Thing**
Although the actions seem coordinated, each company is merging different things. It's like six people moving houses: some are moving furniture, some are transferring household registration, and some are demolishing the house.
- Anthropic (Claude): They merged the interface.
- Action: They incorporated the previously separate collaboration tools (Cowork) and document tools (Docs, etc.) into Claude’s chat window.
- Logic: Users are too lazy to remember which tab handles what. Since it's all AI, let's talk about everything within the chat window.
- Exception: Claude Code, which is used for coding, remained unchanged because it can already generate revenue on its own and doesn't need to share resources with general chat functions.
- OpenAI (ChatGPT): They changed the user interface.
- Action: They renamed ChatGPT to “Classic” and upgraded the Codex app to the new main entry point for ChatGPT.
- Logic: Previously, it was more for chatting; now it's focused on tasks. The name ChatGPT now represents a work assistant that can help with tasks and coding, not just a chatbot.
- ByteDance (Lark/Doubao): They reorganized structures and permissions.
- Action: They split Lark into two parts: the product team was integrated into Doubao (the AI platform), and the sales and service teams were merged into Huoshan Engine (the cloud service).
- Logic: This is the most significant move. Before, the AI in Lark could only “access” your data; now it directly inherits your permissions, allowing it to read your documents, meetings, and schedules seamlessly without additional authorization.
- Alibaba (Qianwen Office): They merged product lines but didn’t complete the integration.
- Action: They combined QoderWork, Wukong, and MuleRun into Qianwen Office.
- Logic: They’re trying to unify the brand, but the internal structure is still messy. There are separate accounts and membership systems for individual and corporate users, with prices ranging from 19 to 78 yuan, which shows they’re still in the transition phase.
- Tencent (WorkBuddy): They kept a distinct approach.
- Action: They incorporated QClaw into WorkBuddy.
- Logic: Tencent is the only one that didn’t need to merge into the chat interface because WorkBuddy has always been a standalone office agent client, like a dedicated AI-powered office computer that doesn’t rely on WeChat or QQ.
- Baidu (KukuAI): They changed the name and re-launched the product.**
- Action: GenFlow was renamed KukuAI.
- Logic: A typical attempt to redefine the product through a new brand.
2. **The Critical Issue: Why Context Is Crucial?**
This is the core of the article. Context refers to the background information that AI needs to know to perform tasks effectively (such as your file content, previous conversations, and permissions).
- Value Source: The more context AI has, the more accurately it can perform tasks and act like a “digital colleague.”
- Cost Challenge: Every time you ask AI to do something, it has to reprocess all the relevant information.
- Data Example: A user test showed that using AI for two days cost 287 yuan in computing power, which amounts to 4,300 yuan per month. 95% of this cost is for reprocessing context, and only 5% is for creating new content.
- Conclusion: If context is fragmented (stored in different apps), AI has to transfer data repeatedly, making it slow and expensive. Therefore, the giants are merging interfaces to keep context in one place to reduce costs and improve the user experience.
3. **The Timing of the Move: Why All at the Same Time?**
This wasn’t coincidental; it was driven by a report released at the beginning of the year.
- Revenue Growth: Cursor (for coding) and Claude Code generated over a billion dollars in annual revenue within half a year, proving that AI Agents can be profitable, especially in the office setting.
- Experimentation Phase Over: At the beginning of the year, companies were testing various approaches. Now it’s clear that the ability to access work context efficiently is the key competitive factor.
- Consolidation Phase: With this realization, they need to consolidate resources to focus on developing powerful office agents.
4. **Data Truth: Who Is Really Using These Services, and Who Is Just Pretending?**
The data in this section is somewhat misleading:
- Tencent (the Only One with Solid Data):
- WorkBuddy has 6.58 million monthly active users, with an average of 19 uses per user.
- QClaw has 2.25 million monthly active users.
- Interpretation: Tencent is the only company with comparable data and shows high user engagement.
- Alibaba (Looks Good, but Doubtful):
- They claim to have over 30 million users.
- Interpretation: This may be due to channel referrals from DingTalk, not actual demand. There’s no daily active user data or retention figures, and their AI division lost 13.8 billion yuan in the second quarter, indicating they’re still investing heavily.
- ByteDance (Lack of Data):
- There’s no official or third-party data on the number of users after the launch of Doubao Work.
- Rumors suggest 200 million daily active users, but this cannot be verified.
- Interpretation: ByteDance might be holding back significant information or the data doesn’t look promising.
- Industry Overview:
- The overall monthly active users in the AI office market are 102 million, with a 8.2% year-on-year increase.
- Interpretation: The market is growing, but the growth rate has slowed down. The competition is shifting from acquiring users to retaining them and generating revenue.
5. **Future Predictions: Four Key Areas to Watch**
If you follow this industry, pay attention to these four indicators in the next six months; they will reveal who will succeed:
1. Task Independence: Can the AI continue to work in the background even when you close the app (e.g., OpenAI’s ChatGPT)?
- Current Status: OpenAI has achieved this; other companies haven’t, so their mergers may just be superficial.
2. When Will Real Data Be Revealed?
- When will Alibaba and ByteDance disclose daily active users and payment rates? If they don’t, it means they’re not ready yet.
3. Cost-Effectiveness: Can the prices cover the costs? Microsoft’s Copilot has a payment rate of 3.3%, and WPS has about 6.9%.
- Interpretation: Few people are willing to pay extra for office software. If AI office solutions are priced high (like Alibaba’s 78 yuan/month), it will be difficult to cover the high computing costs.
4. Permission Model: Will companies allow AI to directly inherit user permissions (like ByteDance’s model) or require separate authorizations (like traditional models)?
- Interpretation: An “inheritance” model provides a better experience but poses security risks; a “call” model is safer but less user-friendly. The company that balances these two will win the corporate market.
---
💡 Insights for Everyone
1. Don’t Be Misled by Mergers: Companies may merge products to save on computing costs, not necessarily for your benefit.
2. AI Office Solutions Are Still in the Loss-making Stage: Except for Tencent, other AI office products are likely losing money. As a business owner, be cautious when choosing a solution and consider its cost structure.
3. Context Is the New Barrier to Entry: Future AI competition will focus on how well products understand your work processes. The platform with the most relevant data will have the upper hand.
4. Be Vigilant: As AI starts to inherit your permissions, data privacy and security will become major concerns.
In summary: This round of mergers marks the transition of AI from a chat tool to a productivity tool. The company that manages context best will emerge as the winner in the next phase of competition.