虎嗅

The Second Confrontation between DouBao Mobile and the SuperApp Barrier

原文:豆包手机与超级App围墙的第二次交锋

Hello! I'm your financial news analysis assistant. This news article about the "Nubi NaviX Ultra" (the second generation of the "DouBao phone") is not just about a new smartphone; it's also about a power struggle on the internet regarding "who really makes the rules."

To help you understand the underlying issues more easily, I'll first summarize the key points and then break down the situation from five different perspectives.

📝 Key Points Summary

In simple terms, nine months ago, the first generation of the DouBao phone was banned by giants like WeChat and Taobao because it was too "domineering"—it directly controlled the phone's screen to place orders or browse WeChat on behalf of the user, which was seen as a violation of their terms of service and led to its functionality being shut down.

Now, the second generation has adopted a different approach. Instead of forcing its way in, it has learned to "knock on the door" and propose a new set of protocols (SAEP and MCP/A2A), which essentially say: "App owners, if you allow me access to certain functions, I can help users with those tasks; if not, I won't interfere."

Although the technology is more compliant with industry standards, the reality is still harsh: currently, only ByteDance's own apps (such as TikTok and Jietou) are cooperating. Competitors like WeChat and Taobao remain closed off. Behind this is a battle for control over the "next generation of mobile interfaces"—should users continue to open apps one by one, or should AI assistants become the sole "butlers" for all tasks?

---

🔍 In-Depth Analysis: Five Aspects of the "AI Phone" War

1. The Shift from "Illegal Hacks" to "Legitimate Cooperation": A Major Change in Technical Approach

Simple Explanation:

Imagine you want to eat at a restaurant.

  • The first generation of the DouBao phone: It didn't ask for service or look at the menu; it directly accessed the kitchen, cooked food, and even tried to steal the owner's account books. Although it was fast, the owner would call the police (as it used high-level system permissions), and the phone was banned.
  • The second generation of the DouBao phone: It now acts like a legitimate waiter. It asks the owner, "The customer wants braised pork—can you provide an interface to place the order?" If the owner agrees, it uses the official MCP/A2A protocol to make the request; if not, it stays out of that app.

Key Point: This change makes the technology more stable and faster, but it also means giving up some control. AI can no longer do whatever it wants; it has to follow the app's rules. For example, if Ctrip hasn't signed a cooperation agreement with DouBao, the AI can't help you book a flight, even though it has the technical capability to do so.

2. The "Moat" of Super Apps: Why WeChat and Taobao are Resistant to AI?

Simple Explanation:

Why are giants like WeChat, Taobao, and Alipay so wary of AI assistants? Because data is their lifeline:

  • Traffic Distribution: Previously, you had to open Taobao, browse the homepage, or search for products. With AI helping with comparisons and orders, Taobao loses control over user traffic and advertising revenue.
  • Data Security: Ma Huateng made it clear that recording the screen and sending it to the cloud is unsafe. For banking and payment apps, allowing a third-party AI to read the screen and simulate clicks poses significant risks.

Current Situation: WeChat only cooperates with phone manufacturers like Huawei, Xiaomi, and OPPO, allowing system-level AI to make calls or send messages. ByteDance is not included because these manufacturers build the infrastructure and don't directly compete with WeChat. However, ByteDance has both TikTok (which attracts users) and an e-commerce platform, making it a rival to Tencent. Therefore, WeChat is unlikely to open core functions to competing AI.

3. The SAEP Protocol: A Test of "Rule-Making Power"

Simple Explanation:

The SAEP (Screen Automation Operation Statement Protocol) issued by ByteDance is about setting rules.

  • Rules: Apps can decide what AI can do, such as checking orders but not posting to social media or deleting chat records.
  • 30-Day Publicity Period: This is a clever strategy. For the first 30 days, only ByteDance's apps and those that explicitly agree are affected. After 30 days, if other apps don't object, some low-risk functions are automatically accessible.

Deeper Meaning: ByteDance aims to become a standard setter in the industry. If these rules are accepted, it will gain significant influence. It's like Apple setting iOS rules; now, ByteDance wants to define the rules for the "AI era." Although only its own apps are cooperating for now, this move is significant.

4. The Battle for Control: Who Will Be the User's "Primary Butler"?

Simple Explanation:

For the past decade, the competition has been about operating systems (iOS vs Android). In the next decade, it will be about AI assistants.

  • Traditional Model: User wants something → Opens an app → Finds the function → Executes the action.
  • AI Model: User tells the AI → AI coordinates apps to complete the task.

If AI becomes the primary Butler, the company that controls the AI will control traffic.

  • Apple: Integrates AI into iOS, making Siri more intelligent to lock users within its ecosystem.
  • Huawei/Xiaomi/OPPO/vivo: Develop system-level AI to connect devices (lights, cars, phones) for a connected home experience.
  • ByteDance: Doesn't produce hardware or rewrite the system; instead, it aims to create an "AI layer" on top of Android. Its goal is to make its assistant the default system-level AI, standing ahead of all apps.

Cruel Reality: Except for ByteDance's apps, other major apps (WeChat, Taobao, Meituan) don't allow the DouBao phone to perform tasks on their behalf. This means the "all-in-one assistant" experience is still limited to ByteDance's ecosystem.

5. Market Outlook: 2026 as the "Year of Intelligent Devices"?

Simple Explanation:

Although the second-generation DouBao phone is currently in an awkward position, the market trend is irreversible.

  • Predictions: By 2026, 45% of new phones worldwide will have generative AI capabilities, with China having a higher proportion of 53%.
  • Competitive Landscape:
  • Apple: Most closed-off, with the best experience but a separate ecosystem.
  • Huawei/Xiaomi/OPPO/vivo: System-level AI, focusing on hardware integration.
  • ByteDance: Application-layer AI, emphasizing cross-app services.

Conclusion:

This battle has just begun. The first generation of the DouBao phone showed how high the barriers to entry are; the second generation demonstrates how difficult negotiations can be.

For consumers, the disruptive impact of AI phones may not be immediate as apps are still defensive. However, in the long run, "service at your fingertips" will become the norm. Future phones might not require you to remember which apps to use; you could simply say to your phone, "Book a flight to Shanghai tomorrow and check the reviews of that restaurant."

The company that can make this possible smoothly will win the next generation of mobile technology. Currently, phone manufacturers (like Huawei and Xiaomi) have an advantage due to their control over the system. ByteDance, on the other hand, needs to rely on its strong content ecosystem and negotiation skills to break through these barriers.