WeChat has quietly added three new AI entry points to its native AI assistant, "Xiaowei," during an ongoing beta test, bringing the total number of AI access points within the app to 16. These additions focus on high-frequency usage scenarios, aiming to streamline how users interact with content and images.
One of the new features is an AI summary tool for official account feeds. On the message page, the "Frequently Viewed Accounts" section now displays a Xiaowei icon and an "AI Summary" button at the far left. Tapping it triggers Xiaowei to compile recent updates from frequently viewed accounts into a digest, allowing users to quickly scan for core information without opening each post. Additionally, an "AI Summary" option has been added at the bottom of individual article detail pages, providing instant summaries of the current article.
For Moments, WeChat has introduced an AI comment feature. Users can long-press on text content in a friend's Moments post to activate an "AI Comment" button. Xiaowei then analyzes the content and generates several comment suggestions in different styles for the user to choose from, simplifying the engagement process.
Image processing within chats has also been enhanced. When viewing an image in a chat interface, a new "AI Process" option appears at the bottom. This allows users to beautify images, extract text information, or make modifications directly within WeChat, eliminating the need for external tools.
Beyond these three primary entries, WeChat is also quietly testing other AI functionalities in limited trials. For instance, when posting to Moments, after uploading a photo, an "AI Write" button appears in the lower right corner. Xiaowei first interprets the image content and then generates multiple draft captions at once. Users can select one, and it is automatically filled into the posting interface—no copy-paste required.
The "Scan" feature has gained a "Photo to AI" visual question-answering capability. After taking a photo, Xiaowei identifies the content and offers follow-up questions, such as brand recognition. For packaging with foreign text, options like "Translate," "Extract Text," and "Summarize Image Content" appear at the bottom, enabling users to perform identification, translation, or further inquiries within a single conversation.
Finally, the chat input box has a new "Command" feature. By pressing and holding the input box to speak, a "Command" option appears at the bottom. Users can slide their finger to that area and state their request. Upon release, the AI interprets the intent and generates a message tailored to the current chat context, which can be sent directly. This function supports polishing expressions, translation, and drafting on behalf of the user.
Currently, all these new features are in the gray-scale testing phase, visible only to a randomly selected group of users. Whether they will be fully rolled out to everyone depends on the official version release.