AI Personal Learning
and practical guidance
Beanbag Marscode1

The AI Search Breakout Battle of Xiaohongshu "Dot Dot Dot": Can Access to DeepSeek-R1 Pry the Search Market?

Under the wave of AI technology, Little Red Book has also accelerated its layout in the field of artificial intelligence. Recently, Xiaohongshu's independent AI search app "Dot Dot Dot" launched a "deep thinking" function. According to industry media reports, the probability is that the function is backed by the domestic AI startup company DeepSeek (search in depth) provided by DeepSeek-R1 Open Source Model Driver.


In the context of WeChat, Jitterbug and other super Apps that are building a closed-loop ecosystem of content through "on-site AI search", Xiaohongshu's choice to cooperate with DeepSeek-R1 is undoubtedly a hope that Rapidly make up for their shortcomings in technology and user experience in a short period of time The company's AI modeling capabilities enabled "Dot" to quickly enter the AI search track and gain key market share. With the power of external AI models, "点点" was able to quickly enter the AI search track and gain a key market share. However, in the long run, it's more like a "Leveraging development" and "autonomous innovation" A balancing strategy between.

The strategic intent of the "Tap" App: Why Little Red Book needs a standalone AI search product.

Why did Xiaohongshu launch an independent AI search app? What kind of search capability has been improved after "点点" is connected to DeepSeek-R1? How is the actual user experience? What are the differentiated advantages of "点点" compared with the common star AI search products on the market? Can "点点" really support Xiaohongshu's ambition in the field of AI search? With these questions, we take the high-frequency search scenarios commonly used by users as a starting point, and conduct a comprehensive review of the "点点" App.

The core search experience of "点点":Scenario-based search is the highlight of life.

As a continuation of Xiaohongshu's slogan "Your Life Guide", "点点" is positioned as a "life scene search". This App mainly aggregates and analyzes the massive note content on the Xiaohongshu platform to provide users with graphic and textual search results to meet the information needs of users in life services and consumer decision-making scenarios.

Unlike the functions of "Da Vinci", "Search Potato" and "Ask Dot Dot Dot" that are built into the Xiaohongshu App, "Dot Dot Dot " is offered as a standalone App and WeChat small program. Since its launch in August 2024, "点点" has been downloaded 200,000 times from the AppStore.

It is worth noting that Xiaohongshu has also previously launched the AI search function "Da Vinci". Initially, "Da Vinci" was also able to search the content of the Little Red Book station for questions raised by users, generate AI summaries, and recommend relevant notes. But from the current situation, "Da Vinci" seems to have been abandoned by Little Red Book. Not only has the official Xiaohongshu account for "Da Vinci" stopped updating, but the function has also failed to effectively respond to the needs of users for integrating notes, generating AI summaries, and assisting in content creation. When asked about the difference between "Dot" and "Da Vinci", "Da Vinci" replied intriguingly, "Dot is responsible for taking care of the search, I'm here to chat with you."

Da Vinci September 2024 Quiz Effects

Da Vinci February 2025 Quiz Effects
Compared with "Da Vinci" and "Search Potato", Little Red Book has obviously increased the promotion of "点点" App. Many Xiaohongshu users have swiped the ads of "点点" in the information stream. Its core selling point is mainly centered on "deep thinking technology + Little Red Book life database", from the travel and play strategy, health knowledge, pregnancy encyclopedia and other specific scenarios of demand into the direct display of the core value of the "点点" AI search.

Next, we will deeply experience the actual search effect of "Dot" App after accessing the DeepSeek-R1 model.

When you open the home page of "点点" App, you can see some scenario-based preset questions, covering a variety of life areas such as eating, drinking, playing, pothole avoidance guide, travel and play guide, extreme melon eating, and worrying tree hole.

After authorizing geo-location, users can directly click on the "Eat, Drink, Play and Be Inspired Nearby" portal on the homepage, or enter a question in the search box to quickly get localized search results.

In terms of the presentation of search results, the "deep thinking" mode of "点点" is very similar to the experience of the DeepSeek-R1 model itself, which shows the user the thinking process of the AI. The substantial improvement of "Dot Dot" in the search experience is mainly reflected in two aspects: Search Intent Recognition Comprehension cap (a poem) multicast dialogue capability The

For more complex or refined search requests, such as "photo friendly and pet friendly Shanghai cafes", "点点" can more accurately understand the user's search intent. At the same time, DotDot also supports multi-language searches in a mix of English and Chinese, such as "date-friendly Shanghai bistro recommendations". ("date" means "date" in English, and "bistro" refers to a small restaurant, and this kind of mixed expression is very popular among young people in Shanghai). This mix of Chinese and English is very popular among young people in Shanghai)

"点点" also supports context-based coherent interactive search. For example, after searching for restaurant information, users can continuously follow up with "How much per person?" , "Is parking convenient?" and "Nearby citywalk routes", and so on.

Keywords with dotted lines in the search results are interactive, and users can click on them to jump to "You May Want to Ask" and related search pages, making it easy for users to further explore content of interest.

However, we noticed that the sorting of notes on the "Dot" related search page is based on semantic relevance rather than the number of likes. This implies that "Dot" focuses more on providing users with relevance and depth of information, rather than simply pursuing popular content.

In the travel planning scenario, "点点" promotion within the Little Red Book App focuses on " 'planning personality' users out of the house magic tool" (The original text is "P people go out god tool", here for easier understanding, modified to " 'plan type personality' user go out god tool")This positioning is in line with the ability of "Dot" to integrate fragmented personal experience into a structured guide. This positioning fits with the ability of "Dot" to integrate fragmented personal experience to generate a structured, illustrated how-to guide.

For example, when we enter "Spain 7 days tour guide", "点点" gives results covering the attractions, ticket prices, camera positions, food recommendations, transportation and travel advice, dress advice, budget reference and anti-theft tips and many other details. The wording and language style of the search results are also more "Xiaohongshuized", close to the daily expression habits of platform users.

Let "Dot Dot" recommend cleansing products for combination oily skin, and the results are also more comprehensive and objective, recommending both high-end brands and affordable products. For users who have difficulty in choosing, the recommendations given by "Dot Dot" can greatly shorten the decision-making time.

When searching for restaurant information, "点点" will simultaneously present the restaurant's actual pictures, price range and special dishes, helping users quickly understand the basic situation of the restaurant.

However, some of the search results of "点点" still inevitably suffer from model illusion (AI-generated content does not match the facts), and mechanical transportation of Xiaohongshu content. In addition, DotDot's multimodal (e.g., image and video understanding) capabilities could be improved.

Some users think that since the content searched by "点点" comes from the notes of Xiaohongshu, the function of "点点" can be integrated into the search interface of Xiaohongshu App, so there is no need to develop a separate App. In this regard, some users have shared their experience that when they search content in Xiaohongshu App, they are easily distracted by other content in the App, which leads to reduced search efficiency. shared their own experience, that when they search for content in the Little Red Book App, they are easily distracted by other content in the App, resulting in lower search efficiency. A pure search App can help users focus more on the search itself and improve efficiency.

These two views actually reflect two different user mental models:

  • Instrumental Mind : When users use "dot" to search, they usually have a clear "problem-solving" goal, which requires a higher level of search efficiency.
  • Community-based mindfulness The user often has the psychological expectation of "browsing" when searching in the Little Red Book App, and is more inclined to accept or even enjoy the process of information dissemination, and find new inspiration and interest in the process of browsing various contents.

Overall. "DotDot's AI search advantage is better suited for scenarios that require quick decision-making, helping users to quickly filter information and shorten decision-making time through structured answers. The

(indicates contrast) Searching within the Xiaohongshu App is more focused on leveraging the diverse perspectives and sensory descriptions of the platform's user-generated content (UGC). For example, when a user's friend suggests going to the Yuyuan Lu citywalk on the weekend and finding a cafe, it's much more intuitive to browse through the two columns of images in the Little Red Book App. For example, when a user's friend proposes to go to Yuyuan Lu citywalk (愚园路 citywalk) on the weekend and find a cafe to sit in, it's obviously more intuitive and efficient to browse through the pictures in the double-column stream of the Xiaohongshu App to get the tips.

In addition to the difference in user mindset, standalone Apps and WeChat mini-programs are also able to reach non-Xiaohongshu users through channels such as the app store, expanding the user base.

In addition, an independent App can establish an independent technical architecture and A/B testing system, which is convenient for the technical team to conduct technical trial and error and rapid iteration. If the AI search function is integrated into the Little Red Book App, it may slow down the iteration speed. It is worth mentioning that just yesterday, when the review was conducted, Xiaohongshu formally closed the "Ask Dot Dot Dot" portal in the App and said that it would be upgraded.

However, the user habits and user migration costs of Xiaohongshu are also factors that should not be ignored. Stand-alone products require users to download an additional App or jump to a small program, and the extension of the operation path may lead to a lower than expected utilization rate of "点点".

From "Lifestyle Community" to "Life Decision Portal": Little Red Book's AI Search Ambitions

After its 2024 Series E+ funding round, Little Red Book was valued at $17 billion. AI search is a key part of Little Red Book's "technology-driven growth" strategic narrative.

This is backed up by an impressive set of data: QuestMobile's data shows that in Q4 2024, Xiaohongshu's average daily search volume was nearly 600 million searches, almost half of traditional search giant Baidu's search volume (about 1 billion searches). What's more, 72% of Xiaohongshu's search users' search behavior is strongly correlated with consumer decision-making (e.g., in the areas of food and beverage, beauty, and travel).

The huge user-generated content (UGC) data in the life category and the consumer decision-making mind of users when searching are the core advantages of the layout of AI search in Xiaohongshu. However, the AI search track is already crowded with strong competitors.

These include traditional search giants such as Baidu and 360, as well as Perplexity AI , minaret AI, code-based search tools for developers Devv AI-native companies such as AI, but also non-search products such as WeChat, Jieyin, and Zhihu. In addition, most generative AI dialog assistants, such as ChatGPT The DeepSeek, Kimi AI searching and deep thinking are also built in, as are beanbags and more.

Although Xiaohongshu has set up a large model team since 2023, and launched a series of models of "Little Groundnuts", as well as multiple AI functions such as AI painting tool "Trik", mirror structure, and chat assistant "Da Vinci", compared with industry leaders, Xiaohongshu is still relatively thin in terms of AI technology strength and overall layout. " and other AI functions, but compared with industry leaders, Xiaohongshu is still relatively thin in terms of AI technical strength and overall layout.

As the underlying AI capabilities are gradually covered by generalized models, the application layer must create new dimensions of value. As Min Kerui, CEO of Secret Pagoda Technology, pointed out, "What you could do originally can now be directly covered by the reasoning ability of, for example, R1, so you need to look for new possibilities on top of it." At present, Xiaohongshu has not yet established irreplaceable functional barriers in the "living search" scene.

However, from another perspective, if Xiaohongshu can successfully seize the "life decision" this has not been fully occupied market high ground, the company may really have the opportunity to rewrite the competitive landscape of the search market.

After all, when young people are staring at their phones late at night, struggling to decide which serum to choose, what they need is not just a smarter search engine, but a "digital BFF" who knows more about the ingredients of skincare products, analyzes skin quality assessments, and combines the experience of their "e-BFFs" on the Internet with their "digital BFFs". They may need not just a smarter search engine, but a "digital BFF" who knows more about skincare ingredients, analyzes skin quality assessments, and combines the experience of "e-BFFs" on the Internet. This is precisely what Xiaohongshu is best at, and also the direction of the future development of the "点点" App.

May not be reproduced without permission:Chief AI Sharing Circle " The AI Search Breakout Battle of Xiaohongshu "Dot Dot Dot": Can Access to DeepSeek-R1 Pry the Search Market?
en_USEnglish