WeChat just deployed an AI agent to 1.4 billion users in one update. No beta. No gradual rollout. One day it was not there, and then it was.
The feature is called 小微 (Xiaowei). It lives in the top-left corner of WeChat, where starred messages used to be. And according to Tencent internally, this is WeChat’s largest update since the app launched.
That is a bold claim. WeChat launched in 2011 and has been through a lot. But the reasoning makes sense when you understand what 小微 actually does, and what it means for brands trying to reach Chinese consumers.
What 小微 can do
小微 is not a chatbot. It is an agent. The difference matters.
A chatbot answers questions. An agent takes actions. 小微 can send messages on your behalf, make voice calls, transfer red packets, query your friend circle, search public accounts, find content on Video Channel, set reminders, handle files, and operate WeChat Pay. It can open and navigate mini-programs across dozens of categories: local services, transport, ticketing, logistics, utilities.
You tell it what you need in plain language. It does it inside WeChat. You never leave the app.
The underlying model is WeLM, WeChat’s proprietary large language model. There are two versions: a 258B parameter model for complex tasks and a 80B model for standard operations. DeepSeek runs as a backup. The system has access to your last two days of chat history for context.
The constraint that protects users (and brands)
小微 operates on a “read-first, act-second” principle. It can access information broadly, but it cannot take bulk actions without user confirmation. No mass messaging. No scheduled sends. No multi-recipient operations. Every financial transaction requires explicit user approval before execution.
This is deliberate. Tencent is moving carefully because every interaction now costs real infrastructure money. In the internet era, serving one more user was nearly free. With an AI agent running inference on every request, the economics are completely different. That is why 小微 defaults to “read mode” and requires confirmation for anything consequential.
For brands, this constraint is actually useful. It means 小微 cannot be weaponized for spam at scale. User actions remain intentional. The quality of any interaction that involves your brand, your mini-program, or your official account content is higher because the user chose it rather than being auto-enrolled.
What this changes for brand strategy on WeChat
小微 changes the discovery layer inside WeChat. When a user asks it “find me a good skincare brand for dry skin in winter,” 小微 will search official accounts, product feeds, and mini-programs to surface an answer. This is a new discovery pathway that did not exist before.
Your brand’s searchability inside WeChat just became a competitive factor. The content you publish in your official account, the product information in your mini-program, the labels and descriptions in your WeChat Shop listings, all of these are now inputs for 小微 to parse when answering user queries.
Think of it as Generative Engine Optimization inside WeChat. The same logic I described in our piece on GEO in China for external AI platforms now applies inside WeChat itself. Your content needs to be structured, specific, and query-answerable, not just visually appealing.
Mini-programs just became more important
小微 can launch and navigate mini-programs. A user can say “order me a coffee from the nearest Luckin” and 小微 will open the Luckin mini-program, locate the nearest store, and initiate the order. The user confirms. Done.
This creates a direct link between conversational AI and mini-program commerce. Brands with well-structured mini-programs, clear product naming, location data, and intuitive checkout flows will benefit. Brands with poorly built mini-programs will be a poor 小微 experience, and the user will try somewhere else.
If you do not have a WeChat mini-program, build one. If you have one that was built in 2020 and never updated, review it. The mini-program is about to become a primary interface for AI-assisted commerce inside WeChat.
The distribution advantage WeChat now has
Every other AI assistant in China, DeepSeek, Doubao, Baidu ERNIE, had to build a user base from zero. 小微 launched to 1.4 billion daily active users on day one. No competing AI product in the world had that kind of instant distribution.
This matters because it shapes the AI landscape in China. Tencent’s Yuanbao AI already had exclusive access to WeChat content. 小微 goes further: it is embedded in the core WeChat experience, not a separate app. Chinese users will interact with AI through WeChat before they think to open a standalone AI product.
The implication for brands: WeChat is no longer just a social media channel. It is becoming an AI operating system for daily life in China. Your presence inside WeChat, your official account, your mini-program, your WeChat Shop, is now also your presence in China’s most-used AI interface.
What to do now
Audit your WeChat mini-program for discoverability. Product names, categories, descriptions, and location data should all be clear and specific. Think about how a user might ask 小微 to find your product, and make sure your mini-program content answers that query.
Review your official account content for search intent. The articles and posts you publish in your WeChat Official Account are now indexed by 小微. Posts that answer specific questions (how to use this product, what is the difference between X and Y, when should I use this) perform better in AI-assisted discovery than generic brand content.
If you are not yet active on WeChat at all, the arrival of 小微 makes the platform more strategically important, not less. A brand with no WeChat presence has no voice in 小微’s answers to its potential customers.
For more on the 小微 launch details, see the Digitaling analysis of WeChat’s AI update (Chinese). For context on China’s AI search landscape, see iiMedia’s AI market report.
If you need to build or upgrade your WeChat presence to be ready for the 小微 era, GMA’s WeChat agency team handles official accounts, mini-programs, and WeChat Shop. Talk to us about your current setup and where the gaps are.
