Summary of Key Points
WeChat has finally collaborated with smartphone manufacturers such as Huawei and Xiaomi to introduce an A2A (Agent-to-Agent) assistant feature. In simple terms, this means that the AI assistants on these devices (like Huawei's YOYO and Xiaomi's Xiaoai) can directly interact with WeChat to send messages or make audio/video calls, eliminating the need to rely on screen reading and button-pressing techniques (OCR simulation). This approach is both secure and balances the interests of all parties involved. It represents a significant advancement in the development of intelligent agents, but it also raises concerns regarding the distribution of computing resources and potential changes in human behavior.
What is A2A, and how does it differ from previous AI interactions with WeChat?
Previously, smartphone AI assistants had to use OCR (Optical Character Recognition) to read the WeChat interface and then simulate button clicks. For example, when Xiaoai was used with apps not developed by HarmonyOS, it would first identify a friend's profile picture in the chat list and then click on it to send a message, which was slow and prone to errors (such as misidentifying the image). With A2A, the AI assistants can communicate directly with WeChat’s own intelligent systems without needing to see the screen or simulate button actions. For instance, you could say, “YOYO, send a message to mom saying we’re coming home for dinner tonight,” and YOYO would instruct WeChat’s internal system to do so, ensuring a faster and more reliable process without revealing any other information on the screen.
Why have major companies chosen A2A?
The reason A2A has become a popular option for both WeChat and smartphone manufacturers is that it resolves several key issues:
1. Data Security: Without simulating button clicks, no private data (such as chat history) is exposed, providing peace of mind for both WeChat and users.
2. User Retention: Smartphone manufacturers’ AI assistants make their own apps more attractive to users, while WeChat can retain users by integrating these assistants, preventing them from being used by third-party tools.
3. Cost Allocation: Communication between intelligent agents incurs “token fees” (similar to service charges for AI conversations). With A2A, the costs can be shared between WeChat and manufacturers, avoiding a single entity bearing all the expenses.
Compared to the previous OCR-based methods (which were more prone to privacy breaches), A2A is a more balanced solution that benefits everyone involved.
What possibilities does A2A hold in the future?
While current A2A functionality is limited to sending text and making audio/video calls, there are significant potential expansions:
- WeChat’s Ecosystem: Mini-programs within WeChat could be integrated with A2A. For example, you could ask YOYO to use a Meituan mini-program to order tea.
- Cross-app Interaction: A2A is a versatile technology that could enable commands like “Xiaobu, check last year’s travel report using Ctrip” or “Xiaoai, pay the electricity bill with Alipay.”
- International Adoption: Google’s Gemini Spark already supports A2A, and Apple is expected to integrate it into Siri at its WWDC 2026 event. ByteDance and ZTE’s new smartphones have also shifted from OCR to A2A.
In the future, it’s possible to have a single AI assistant that can handle all app functions.
Challenges with A2A
Despite its advantages, there are two important issues:
1. Costs: Who will cover the costs of these token fees? Each interaction between intelligent agents incurs expenses, which needs to be determined among users, manufacturers, and app developers. Whether the service will be free or paid remains an unresolved challenge.
2. Human Interaction: If everything is handled by AI, people may start to express themselves in more structured, machine-like ways. Over time, this could lead to a loss of natural, flexible communication skills.
However, there’s no need to worry too much for now; WeChat’s A2A feature is still in the testing phase and will likely take at least half a year to cover all daily tasks. The ability to simply use voice commands to interact with WeChat is already a convenient advancement brought by technology.
In summary, A2A marks a crucial step towards making intelligent agents more practical. It addresses real-world issues while balancing various interests. While future costs and ethical considerations need further exploration, for ordinary users, the ability to communicate more effortlessly through voice commands is certainly a welcome development.