Article by Sleepy
By the end of 2022, not long after ChatGPT was released, Ma Wen Chao borrowed an employee's phone. He typed a question in the chat box: Will Xiaohongshu (RED) be disrupted?
It was reported that since then, he asked the team to report on AI progress every two weeks. Every fortnight meant the machine had not provided him with a reassuring answer.
In August 2023, he wrote in an internal memo that he discovered while chatting with foreign friends that many questions people asked on ChatGPT were about life experiences, such as how to choose a product, how to use it, and how to avoid pitfalls, which overlapped with Xiaohongshu's business.
However, he went on to say that this was because there was no such experiential accumulation overseas, while Xiaohongshu had it. This moat was still difficult to be shaken by AI for the time being.
The term "moat" was previously something entrepreneurs said to impress investors, but this time it seemed like he was saying it to his anxious self.
That year, Xiaohongshu had just turned ten, with a monthly active user base exceeding 300 million, turning its first profit, with a revenue of $3.7 billion, a net profit of $500 million, and the expected doubling of profits the following year, surpassing $1 billion.
In business history, companies die in two ways, either from poverty or from wealth. Those who die in poverty are countless and not worth mentioning. Those who die in wealth always make the news, like Kodak had money when it died, and Nokia was still the industry leader when it perished.
Having lots of money and a long life are two different things. Wealth does not exempt one from fear; it only turns fear into specific actions.
In 2026, these actions intensified.
On June 8, RED Skill was launched on Xiaohongshu, where a component could be attached below a note, copied to the Agent, and used.
Going back further, on April 30, the AI division, Dots, was established, integrating models, infrastructure, and engineering products, reporting directly to the new CEO, Conan.
Even earlier, it acquired the development company of an AI search product, along with obtaining a payment license.
On the strategic investment list, MiniMax, Dark Side of the Moon, and a string of AI hardware companies began to appear.
Over the past thirteen years, the consumption experiences, lifestyle habits, and daily judgments left by hundreds of millions of users in notes were its true foundation. With AI's arrival, it aimed to reprocess these judgments, first turning them into answers, then into tools, and finally into business. To avoid being disrupted, it had to take the initiative.
Is experience something that can withstand processing? To answer this question, we need to go back to 2013, back to China's own Age of Exploration.
The Age of 70 Million People's Exploration
In June 2013, Qufan Fang left his job at a foreign company and, together with Mao Wenchao, founded Xiaohongshu in Shanghai. Their first product was not an app, but a PDF titled "Xiaohongshu Outbound Shopping Guide."
That year, the number of outbound Chinese tourists exceeded 70 million, equivalent to the entire population of France taking a trip abroad.
While Europeans' Age of Exploration brought back spices, gold, and colonies, the Chinese Age of Exploration brought back cosmetics, rice cookers, and guides. Despite the small items, the intention was the same—to bring back good things from afar.
The world of goods outside the country's borders suddenly opened up, with duty-free shop shelves crowded with tourists holding smartphones, none of whom were told what was worth buying. Information asymmetry is like a mineral deposit; whoever first gathers the experiences of those who have been there can become the mine owner.
The PDF was posted online and was downloaded 500,000 times in less than a month. A few months later, it evolved into an app, and a few years later, it found its way onto hundreds of millions of phones.
When Chinese people encounter something, they never ask for a manual; they ask a person.
Fei Xiaotong wrote in "From the Soil of the Country" that trust in rural society does not rely on contracts but on familiarity. Those learning a craft ask their masters, new daughters-in-law ask their mothers-in-law, and those venturing into the city for the first time seek out fellow villagers. For thousands of years, experience has been passed down from generation to generation—slowly but sufficiently.
Sufficiency has two prerequisites: people living close by and time moving slowly. These two prerequisites were breached in the past few decades. Hundreds of millions of people left their hometowns to live in buildings where they know nothing about their neighbors. The range of purchasable items expanded from a few hundred on cooperative store shelves to billions on e-commerce websites. It's challenging to ask an elder who has never used a robotic vacuum cleaner which model one should buy. The experienced individuals have not had a chance to share their experiences.
The internet claimed it would solve this problem, but it ended up exacerbating the issue. People invented the internet to access information, but eventually, there was so much information that no one dared to trust any of it. Most online information comes from sellers, and sellers' job is not to help you make judgments but to persuade you to pay. Judgment can only come from those who don't profit from you.
Xiaohongshu consolidated the scattered "I've tried it" experiences among hundreds of millions of strangers. A girl from Guangzhou wrote that a certain foundation would cake on her oily skin, a young man from Shenyang listed eleven pitfalls he encountered during renovation, and a mom hesitated for days between two types of baby food wrote about her experience.
Most of the people who write these notes are unknown, not experts, their writing is not precise, and may contain personal preferences and errors. However, these words carry warmth.
Encyclopedias aim for definitions, advertisements aim for persuasion, but these notes aim for nothing. They are just testimony, imperfect testimony. In a courtroom, the most credible testimony is often this way. Overly perfect testimony is like it has been memorized. Later, the industry gave a name to this phenomenon, "grass planting."
By the end of 2024, this App's daily search volume will approach six billion times. People here search not so much for knowledge, mostly for life, such as decoration, essentials, and travel guides. While search engines provide you with information, Little Red Book provides you with other people's experiences. Of course, there are advertisements in it, and it may not necessarily give you the most accurate answer, but people are still willing to read because many questions in life do not have standard answers.
Behind six billion searches are six billion hesitations, people holding their phones in the middle of the night unable to make up their minds. This is all that Little Red Book is about.
And then, AI came.
The End of Patience
Thirty years of the Internet is a history of human patience wearing thin.
In the portal era, information was organized into directories, and people had to find it themselves. In the search era, it became links, and people had to click on them themselves. With the advent of the information stream, you don't even need to search; algorithms feed you. Each change has shortened patience by a bit. In the AI generation, information is turned directly into answers, and human patience has reached its limit.
This is not the fault of the users. Human love for convenience is endless; the wheel, the elevator, the remote control were all invented for this purpose. Once a person gets used to an AI dialog box, it is very difficult to go back to the days of searching through threads and filtering on their own.
The challenge of Little Red Book is that its most valuable part is precisely the hardest to compress into a single answer.
In the past, people would read twenty notes here, compare, hesitate, and finally make their own decisions. This process was slow because you could see three people approving, two regretting, and one reminding you that this item is useful but delicate. Someone would write that the hotel had poor sound insulation but great breakfast; this sentence is useful because it comes from a specific person, and you can probably guess what they care about, and then decide whether their experience is valuable to you.
AI is like a pre-made meal factory; what goes in is the variety of life experiences, but what comes out is a standard recipe. While it is convenient, what you can remove—hesitation, failure, and prerequisites—is precisely the most valuable part of the experience.
Experience always grows from specific individuals—skin type, city of residence, budget—all determine whether a suggestion is useful. The answers provided by the machine do not have these prerequisites and sound like slogans. Slogans can't help you choose a foundation.
Xiaohongshu understands this dilemma. Patience cannot be sustained. When the day comes, its six billion searches will become someone else's model corpus, and it will become a mine, an open-pit one at that, where anyone passing by can take a shovel.
So it had to take matters into its own hands. They were not too late to act. Starting in 2023, they independently developed the "Small Sweet Potato" model, launched the AI drawing tool Trik, and conducted internal testing of the conversational product "Da Vinci." Most of these products did not make big waves, but they were not done in vain. They were like rounds of probing, with Xiaohongshu needing to first figure out what AI can actually do for it.
The real pioneer in finding the way is Dots. It conducts life searches, integrating internal notes and web-wide information, allowing queries in both text and voice. Later, Xiaohongshu simply acquired the company behind Dots. Dots may not be an explosive success, but reconnaissance is not meant for storming the castle.
It discovered one thing: in the past, searches started from keywords, and what users received was an address; now, questions start from a situation, and what they receive is a whole set of issues. People are no longer just searching for "Okinawa family trip"; they are asking how to plan a five-day trip to Okinawa with a three-year-old, a budget of fifteen thousand, and a preference for a location closer to the sea.
To address these issues, Xiaohongshu has successively released research on multimodal retrieval and search understanding, open-sourced the image editing model FireRed, and the search agent framework REDSearcher. It has no intention of competing with tech giants on universal models. While others compete on parameters and rankings, what it aims to do is to understand, disassemble, and reassemble the scattered real-life experiences in text, images, videos, and comments into specific and actionable recommendations. This year, Dots was established, and this thread of experimentation from the edge has entered the core business.
Xiaohongshu was willing to do the work of piecing together answers from twenty notes for the users. But one answer can only solve one problem. What it truly wanted was to turn experiences into a capability that could be repeatedly leveraged.
Notes Sprouted Hands and Feet
RED Skill does just that. It transforms experiences from content into tools.
After the feature went live, Xiaohongshu quickly launched support activities and curated lists, with three hundred thousand people starting to write AI Skills. For example, the PPT generation tool developed by Guicang received over ten thousand stars on GitHub and attracted several thousand installations on Xiaohongshu within a few days.
If we look back, the independent development competition last year received 1355 projects. In the spring of this year, the first Hackathon was held, lasting forty-eight hours with a prize pool of five hundred thousand yuan. Sixty percent of the participants were post-2000s, with the youngest being twelve years old. The number of notes on the platform about "Build in Public" has exceeded 1.1 million.
Although these numbers are not yet enough to prove that the ecosystem has taken shape, they are enough to show what Xiaohongshu wants to achieve.
In the past, when developers wanted to kick-start a product, they mostly went to GitHub or Product Hunt. There were many peers and investors there, but not necessarily many regular users. Some would star your project, some would give you valuation, but not necessarily would anyone place an order.
What Xiaohongshu is targeting is precisely this gap. Developers write about their progress here, users provide feedback in the comments, bloggers turn their user experiences into notes, and the platform then uses rankings to attract the initial attention. For an AI tool, just writing it out is only the beginning; it also needs to be tested, discussed, and translated into something understandable and useful for the general public.
Creating tools may not be Xiaohongshu's greatest strength. However, integrating tools into daily life, it is very familiar with.
Over the past thirteen years, Xiaohongshu's creators have been more like storytellers – vivid in their writing, trustworthy in their recommendations, accumulating influence bit by bit. Users are willing to listen to you primarily because they trust you as a person. In the AI era, creators are beginning to transform into craftsmen. From literati to craftsmen—this may sound like a step down, but it's just a change in metrics. The number of people using a tool, the number of times it's invoked, and how much it actually accomplishes for users are starting to determine a creator's worth.
For those who write notes, in the past, your experience could only be seen; now it can also be invoked. If it can be invoked, then there is the possibility of pricing.
Before the Search Term Appears
In December 2024, Dai Lidan, a partner at Today Capital, joined Xiaohongshu as the Chief Strategic Officer to establish a venture capital team. She graduated from Peking University with a degree in Computer Science, worked on Baidu Image Search and Baidu Maps, then went to Harvard for an MBA, and returned to China to join Today Capital. She has experienced technology, product, and capital.
Before her arrival, Xiaohongshu had mostly invested in consumer brands such as M Stand Coffee, Moody contact lenses, as well as food, trendy gadgets, and mother-infant products, focusing on the lifestyle of young people, which was also her expertise. After she arrived, financial investment and strategic investment were separated, and the venture capital team shifted its focus to deep tech and AI. Xiaohongshu is part of MiniMax's shareholder list, and it was also involved in the round of financing that exceeded one billion dollars.
It is not just betting on AI on the screen.
In the area around the Shenzhen Nanshan Science and Technology Park, with DJI headquarters as the center, there is a group of people working on AI hardware. In the second half of 2025, Xiaohongshu invested in nearly ten startups here, moving quickly, sometimes closing a deal in a day or two, and willing to use a higher valuation to grab a share.
Two of the investments were made through its subsidiary, "Potato Can Do Magic." One investment went to Yunwang Innovation, a company that turned a traditional foam roller into an AI massage robot that can sense where the body is sore, adjust the pressure and path accordingly; the other investment went to Skyris, which creates a companion robot that floats in the air with helium, interacts with people using wings, LED eyes, and voice.
Internally, Xiaohongshu (Little Red Book) is affectionately called the "gateway to lifestyle decisions." These eight words look great on a PowerPoint slide, but pleasant words are often far from reality.
Decision-making is already a late stage; when someone starts searching for how to use a foam roller, it means the need has already been expressed. Before it becomes a search term, the need often doesn't have a name yet—it might just be a continuously sore shoulder or someone sitting at home for three hours.
In the past, Xiaohongshu focused on the downstream, waiting for people to turn their life experiences into notes. Now, it wants to move upstream, proactively identifying those needs that have not yet become search terms.
In 2024, Xiaohongshu's parent company also invested in a fund under Kunshan Venture Capital as a limited partner. Kunshan was one of its early investors, discovering the company at a startup competition in 2014 and investing the following year. Ten years later, the recipient of the investment became an investor. Xiaohongshu exchanged a fund share for a long-term channel to early-stage projects.
Of course, investing early does not guarantee accurate predictions. AI hardware has yet to prove its ability to commercialize on a large scale—mass production, the supply chain, after-sales service—all are arduous tasks and not the kind of business Xiaohongshu is familiar with. The more troublesome issue is data. The device knows when your shoulder is sore, but the platform also wants to know why. Knowing too little makes the product unusable; knowing too much poses privacy risks.
However, it still wants to invest. What it truly worries about is not today, but that person in the deep night of tomorrow who won't open Xiaohongshu to review notes but will directly turn to another AI for answers.
When Advertising Moves into Answers
Xiaohongshu's story cannot escape commercialization.
On this platform, experience and business have always been intertwined. Behind skincare advice are skincare products, behind home decoration tips are building material suppliers. Users want to take shortcuts, merchants want to be seen, and the platform wants to make money from it. Each desire individually is reasonable, but together, there must be a set of rules.
In November 2025, Xiaohongshu acquired the Eastern Payment License through a subsidiary, completing the final piece. AI can recommend products and services for users, and after the recommendation, deciding where the order is fulfilled, where the money goes, determines who ultimately benefits from this business. Xiaohongshu doesn't want to only provide recommendations but also wants to keep the transactions within its ecosystem.
Red, a pioneer in commercializing content, made its move as early as December 2024. At the WILL Business Conference, Red unveiled the AIPS Audience Asset Model, utilizing the Grassing Alliance to connect data from Taobao, JD.com, and Vipshop, reconciling it with brands' own data. During the conference, two intriguing figures were shared: the decision-making cycle for facial serums, up to twenty-nine days; and infant supplementary food, exceeding seventy days.
This ambiguity is precisely the enigma of influencing. An individual may read a product review today, check the ingredients ten days later, place an order on another platform twenty days later, watch a livestream in between, ask friends for recommendations, and the ultimate conversion credit remains a mystery. Brands have always been in the dark about this, but AIPS aims to illuminate this fuzzy path.
Red's true value lies not in mere traffic. While someone might casually scroll through short videos for entertainment, once they start researching serums or baby food, a purchase is likely imminent.
The most valuable insight is understanding what people are hesitating about. AI can discern these hesitations more clearly now—platforms used to know merely what you viewed, but they can now decipher what problem you're trying to solve. You are no longer handing over just a keyword; it's a whole scenario including budget, preferences, health status, and those unspoken concerns.
Advertising has always been tapping into human judgment. Initially, it stood on the roadside as a billboard—an obvious ad that one could choose to ignore. Subsequently, it merged into articles as advertorials and product placements. Later, it infiltrated the information flow, blending in more seamlessly with the content you'd naturally consume. With each progression, ads became harder to detect yet closer to influencing decisions. In the era of AI, ads found a new home within the answers.
Machine Learning Masters "I've Tried It"
In February 2026, following the national "Artificial Intelligence-Generated Synthetic Content Identification Measures," Red mandated content creators to label AI-generated graphics, text, and videos; undesignated content faced distribution restrictions. In March, it commenced purging AI-operated accounts entirely managed by machines, leading to direct suspensions. In April, Red unveiled its comprehensive AI governance principles, encouraging AI to magnify creativity while opposing AI's fabrication of life—cloning voices, fabricating personas, and fictionalizing experiences are all prohibited.
While these statements may sound like declarations, they are actually survival tactics.
AI excels at mimicking humans; however, the phrase it learned the quickest and which it really shouldn't have is "I've tried it." For Red, the trust built over thirteen years relies on countless specific "I've tried it" testimonials. Machines can churn out ten thousand product trial notes, but they have never genuinely experienced a product. Their skin never reacts, and their wallets never feel the pinch.
When content of this nature becomes abundant, human experience also depreciates. Little Red Book will once again transform into what it originally sought to replace, becoming a collection of more beautifully written, more human-like seller scripts.
Nothing about the future is certain. Whether RED Skill can establish a true ecosystem, whether DotDot can integrate into the main platform, whether payments will be incorporated into the solution—all of these aspects will be left to time. However, the nature of this matter is already clear: Little Red Book is currently acting as a translator, translating human experience into a structure that machines can process, turning real-life judgments into tools, and intertwining hesitation with business.
Translation emphasizes faithfulness, expressiveness, and elegance, and machines have already learned expressiveness. What Little Red Book must hold onto is faithfulness.
Borges once wrote about an empire obsessed with precision. The art of cartography there became increasingly sophisticated, where a map of a province ended up as large as a city, and a map of the empire as large as a province. The cartographers still found this insufficient and eventually decided to create a map the same size as the empire's territory, where every city, road, and barren land could be identified on the map. But once the map became as large as reality, it became useless. Subsequent generations paid no attention to it, letting it decay in the desert.
AI is currently drawing such a map for experience, becoming more detailed and faster, eventually making it easier for people to forget that the map is not life itself.
In an email, Mao Wenchao mentioned that this moat cannot be easily shaken by AI in the short term. He probably also understands that the real issue is not the moat but the city. Little Red Book must create an increasingly intelligent machine; otherwise, the thirteen years' worth of experience will soon be organized, accessed, and repriced by others. However, once the machine's voice overshadows human voices, the city will be empty, with the moat guarding an empty city—no matter how wide it is.
It needs to integrate the machine into the city while ensuring that what remains in the city is not just the machine but also those indecisive individuals in the late night and the people willing to say "I have tried" to them.
This is its true moat of defense and its current source of unease.
Epilogue
Just before the final draft of this article, Bloomberg reported that Little Red Book plans to secretly submit an IPO application in Hong Kong by the end of this month, with a valuation that once reached $31 billion and an estimated full-year profit of around $3 billion in 2025.
From a simple PDF to the Hong Kong Stock Exchange, thirteen years. It has accumulated the hesitations of hundreds of millions of people's lives into something profitable, and now it is up to the capital markets to reevaluate it.
Stock prices will always fluctuate. However, those individuals who are indecisive in front of their phones late at night, those willing to say "I have tried" to strangers, will not disappear from the story due to stock price fluctuations. Money may make a company run fast, but running for a long time is another matter.
Leave the future to time.
Welcome to join the official BlockBeats community:
Telegram Subscription Group: https://t.me/theblockbeats
Telegram Discussion Group: https://t.me/BlockBeats_App
Official Twitter Account: https://twitter.com/BlockBeatsAsia
