All the social media channels now have in-app AI search, chat, and some even enable users to attach their vibe-coding projects. This changes the targeting audience to include AI as a target to read a brand’s content. Ranking stops being only about good cover image and an enticing title, and starts being about whether your post is specific and complete enough to get quoted. These features are still limited, but they show an inevitable trend for AI adoption in a brand’s China operation.
Red’s mini-tools turn a post into its own Mini Program
Last week, Red launched 小工具, a vibe-coding add-on, as part of Builder Hub, Red’s section for AI creators. It enables content creators to create a tool and attach it to a post. The billboard shows the top weekly user amount is already over 650k users. The 2 most popular categories are mini games and personality quizzes.
 |
 |
| The Top 100 board of creator-built mini-tools, refreshed daily at 8pm. | The casual-games tab of the same board. A BBQ simulator leads on 642,000 users, and six of the top seven are games. |
Since February 2026 creators have been publishing 40k to 50k vibe-coding works a month on Red, with no way to attach one to a note until now. This enables a Red post to connect to its own “Mini Program”.
Although currently it does not support payment. Thus no direct monetization path for a tool other than earning followers and engagement.
While Red is very restrictive on external links (i.e. links to WeChat, personal WeChat account, Tmall or any other external links even in DMs are strictly blocked), the vibe-coding add-on is a perfect way to lead users to a custom-designed tool without redirecting the traffic to other sites.
 |
| A mini-tool in a post footer. 2,716 people have used this “app” |
It requires following Red’s own code template, and uploading it as a packaged zip capped within 30MB.
Expect lots of brands to leverage this feature to create engagement campaigns with Red users. Once users connect with the vibe-coding add-on, brands can then leverage Red ads to retarget the users. Since Red ads are often limited in the tag/customization choices, a viral brand-related game/personality quiz will create user engagement and enrich the audience group.
Red in-App AI assistant ranks the search
73% of Red’s monthly actives use search, and it’s a strategic metric for Red. The press calls it the platform’s 命门, its lifeline, as it keeps users within App. Now Red search has a new AI assistant feature called 点点 (Ask Dot) to rank popular notes.
Red tried this as a separate app first and it stalled. Monthly downloads of the standalone Ask Dot app fell by 55% to only 36k downloads 3 months after its launch in Feb 2025. The AI is now integrated in Red search.
So far Ask Dot has reached tens of millions of daily actives and lifted retention by 2% – 3% generating answers from Red’s own embedding database.
 |
| Ask Dot has its own tab in Red’s search bar, next to Users, Products and Images. |
Most of the notes that show up have a clear product name and brand name, and are usually posts with over 100 likes or recent notes.
 |
 |
| Summarize content from 15 Red posts to define brand and pricing for a brand. | Highly ranked notes carry the brand and product name in the title, and have a high like count. |
High-quality content with product keywords will help with the algorithm in search. The top rank notes in the future will no longer be just a good cover picture and an enticing title. This means brands need to optimize the post content to include the full product name, product selling points, and relevant long-tail keywords and even target audience group or use case occasion in the posts. This will improve the traffic of an old post in the long run.
Also Red officially trained Ask Dot to prioritize genuine discussion content instead of paid media. Brands should still focus on generating real and organic engagement.
 |
| Ask Dot as a standalone chat: 68 posts read from Red and the wider web, then Chanel broken down by category, resale value named. |
AI reads video too
Red AI assistant is beta testing summarizing video too. Red splits a clip into timestamped chapters and writes a one-line description of each. It means scripts are being read as content.
 |
| A 60s tutorial split into timestamped chapters, each with a one-line description, before the viewer watches it. |
Community engagement boost with AI comments
Many users already use 点点 (Ask Dot) in the comments if they have a question about the content of the post. Today most of the comments use Ask Dot as a Wikipedia.
 |
| A creator @Ask Dot to ask why a note from 18 months ago suddenly picked up traffic. |
AI shopping assistant launches with limited capability
Ask Dot also has a shopping assistant feature. Red’s own promotion markets this as people finding goods becomes AI finding goods for people.
 |
| Red’s own promotion for AI shopping assistant: the AI picks the product |
Ask Dot what to use for dry skin to get a shopping list links to Red store. The brand’s official flagship store takes the top slot. The resellers below it are ranked on credibility, such as units sold, cart adds, saves and lifetime store totals, rather than by price. This would be a strong argument for brands to open up the flagship store going forward.
 |
 |
| Ask what to use for dry skin La Mer’s toner comes back at 815 RMB on a buyable card. | Tap that card and a ranked merchant sheet opens, the brand’s own flagship first, units sold and cart adds under every listing. |
Although this is still a testing feature. If we search for other categories such as bags and dress, Red prioritize on pushing offline stores and official websites. It’s not fully integrated to all the categories.
Brands should still treat GEO as a Red discipline.
- Write the product information as if AI is the buyer. Include the product name, model name, target users, the core selling points and the use scenario explicitly.
- Start competing to sit inside the generated answer, stop only posting beautiful cover pic with an aspirational caption.
- Treat store reviews as ranking input. Red has already started putting AI summaries on some stores’ 店铺口碑, which signal feeds recommendation.
Note that due to the competition between Red and Tmall, Ask Dot puts a “friendly” reminder to warn users not to purchase the brand on Taobao to avoid fake products. It’s not recommending Red store as a channel yet, but it’s likely to add this integration soon.
WeChat’s AI assistant is integrating into the whole APP
Red is building AI into search and comments. Tencent is going wider, and putting the AI assistant in front of everything.
WeChat launched its AI assistant 小微 Xiaowei in June this year. It’s now adding more than 12 App entry points within WeChat: search box, top of the WeChat article, share page, and comment, OA message summaries, Moments caption generation, Scan to take photo for AI analysis, the Frequently Read banner of your subscription list and the share sheet when you forward a video. It has even reached group chats, deliberately placed outside the chat stream itself so the conversation stays human.
 |
 |
| WeChat’s search bar is now an entry point to Xiaowei. Hold to talk/type, in English or Chinese. | Brand summary with the collaborations and a price range. |
Currently you are able to ask it to summarize the last WeChat articles you’ve browsed, or the most popular content on your WeChat Moments. It doesn’t connect with e-commerce yet.
 |
 |
| Summarises the WeChat articles you have recently read, and content links. | Summarize what is trending content in your Moments feed |
The summary card renders a small cover, top product display, then links to an official account.
 |
| Xiaowei closes its answer with a verified brand official account card: a new route into a brand’s owned WeChat presence. |
The Full Agentic Shopping Experience
WeChat Pay is already integrated within Xiaowei. Users can set up a spending limit and confirm key steps with AI专属卡. WeChat ecosystem Apps such as Meituan, Didi, Ctrip, JD and Dewu (POIZON) are already integrated with Xiaowei.
Today, you can ask Xiaowei to order Luckin coffee, and it will find the nearest Luckin and complete payment.
It also supports users ordering takeout, although the UX is currently bad. It works like a CLI that doesn’t have a clear interface:
 |
“Order hot and sour soup takeout”
↓
Opens the Meituan mini program
↓
Auto-detects the location, and adds the top-ranked merchant’s product to the cart
|
Still, it’s impressive that it’s able to navigate within the Mini Program, pick a merchant that can deliver to my address and add a product to the cart. Note it picked a large chain merchant, not from my favorite restaurants.
It’s easy to imagine such a user experience can be adapted for WeChat store and e-commerce eventually.
Tencent has included Xiaowei in its 2026 Q2 financial report as a highlight and a focus for the future. Pony Ma framed the quarter as building “an entirely new, AI-empowered Tencent” across intelligence, applications and infrastructure. Thus a network-wide SEO/GEO is going to be more critical for content.
Optimize for the AI, but keep the 活人感
While Red is becoming more AI-driven, it’s still one of the most anti-scripting and anti-automation platforms. It actively blocks automation attempts by LLMs in order to protect its authentic community. So brands that want to leverage AI to run marketing need to be aware of the terms of use to avoid account bans.
Red has handled over a million AI-related violations so far this year, including more than 800k AI-run accounts and close to 150k AI-fabricated notes (according to Tencent News, April 2026). The rules explicitly ban AI-run account operation, AI-generated impersonation, and faked-identity livestreams.
As AI is unavoidably welcomed into every aspect of social media operation, for premium brands, content showing “behind the scenes” or founder vlog is getting stronger traffic. A word that’s getting more mentions these days is 活人感, live-person feeling. Even if brands need to fully adopt AI in operation, content creation requires an even stronger connection by showing the imperfection and work in progress.
Same for WeChat. WeChat bans using AI, scripts or interfaces to impersonate, create or publish content on official accounts. Punishment includes traffic restriction, deletion or even account bans.
Conclusion
While both Red and WeChat are not fully agentic on commerce yet, the trend is to integrate their AI assistants with more data access. Brands need to consider the accessibility of their content to both AI and humans.