Summary of the Key Points
This report clearly explains the contradictory approach taken by REDnote recently: on one hand, it has heavily invested in AI technology; on the other hand, it has taken strict measures to regulate AI-generated content. Over the past three years, REDnote has gone from believing that its foundation of real-life user experiences was unassailable to experiencing a decline in user engagement in 2025, with AI-native apps like DeepSeek stealing user time. For the first time, REDnote has elevated AI to a strategic level on par with its community, e-commerce, and commercialization efforts. It has merged its two AI products into a single AI search service and even open-sourced its large models to showcase its technological capabilities. At the same time, REDnote has taken drastic action against over 150,000 AI-related violations and thousands of accounts, making it the domestic platform with the strictest control over AI content.
The underlying logic is practical: REDnote is using AI to maintain its core advertising revenue and to boost its valuation in preparation for a potential listing on the Hong Kong stock market. However, it cannot afford for AI to undermine the trust in its community, which is built on real, human-generated content. REDnote has chosen a different path from giants like ByteDance and Baidu, which focus on developing general-purpose large models. Yet, it has not yet solved the dilemma of how to increase advertising revenue without reducing the usefulness of its AI search function.
---
Detailed Analysis
1. Three Years of AI Development: From Indifferent Observation to a Rush to Invest
In the first two years of its AI efforts, REDnote was merely experimenting on a small scale. After ChatGPT became popular in 2023, founder Mao Wenchao believed that overseas users' interest in AI-related questions was due to the lack of a vast database of real-life experiences on REDnote, and that AI would not threaten its position. The AI-powered painting and text-to-image features were seen as minor supplementary tools, not as strategic projects. The turning point came in 2025 when DeepSeek's daily active users reached 30 million, and REDnote observed a significant change in user behavior: instead of creating their own travel or product guides by browsing multiple notes, users were asking for pre-made lists from AI and then verifying details on REDnote, significantly reducing their time spent on the platform. Data from Nomura Securities confirmed this trend, showing a 1% decrease in total usage time and a 4% drop in average daily usage per user in May 2025—the first negative growth since data tracking began. In contrast, DouBao's daily active users increased by 3.6 times to 158 million. This shift forced REDnote to elevate AI to a primary strategic focus and invest heavily in its own AI products.
2. Focusing on AI Search to Protect Its Core Business
Many wonder why REDnote would invest in AI rather than focusing on its community. The reason is simple: 76% of its revenue comes from advertising. The longer users spend on the platform, the more ads they see, and the more advertisers are willing to pay. If users switch to other AI services, REDnote would lose this revenue source. However, REDnote faces a paradox: while better AI search makes ads less effective, it also reduces ad space. Additionally, the Hong Kong stock market values companies with AI capabilities, even if they don't generate immediate profits. Therefore, REDnote must showcase its AI efforts to improve its valuation.
3. A Contradictory Approach to AI Content Regulation
While other platforms encourage users to create AI-generated content, REDnote takes a strict stance. It prohibits AI-generated faces, fake notes, and bulk account creation, as these go against its core value of real user experiences. Its unique asset is its algorithmic recommendation system, which ensures that users see trustworthy content. If AI were to create random, misleading content, users would lose trust in the platform. REDnote's approach is to use AI to organize existing real notes into useful summaries and guides, maintaining its core strength.
4. A Differentiated Approach from Big Tech
Unlike ByteDance, Alibaba, and Baidu, which focus on developing general-purpose large models, REDnote specializes in AI that understands user needs in specific areas of life. It has hired experts in philosophy, literature, and anthropology to train its models to understand the community's tone. While it may not compete with these companies in complex tasks, it excels in providing practical advice, such as recommendations for weekend activities with children or safe dining options. This differentiation helps REDnote avoid direct competition in the general AI field.
5. Facing Challenges in Revenue Generation
REDnote has completed the initial steps of its AI strategy: it can integrate large models into its products and provide more convenient search results, increasing user retention by 2-3%. However, it has yet to find a viable revenue model. Three attempts at AI-based monetization have failed: traditional ads within AI results have low click-through rates, and direct product recommendations by AI have undermined user trust. The third approach, where AI directs users to third-party shopping platforms for commissions, also results in significant profit losses for REDnote. Brands are reluctant to invest in REDnote's AI search due to the lack of a separate ad market. REDnote's challenge is to create a new advertising model that maintains user trust and generates revenue without damaging its community.
In summary, REDnote's approach to AI is both strategic and practical, aimed at protecting its core business while adapting to market trends. Although it faces challenges in generating revenue, its focus on real user experiences gives it a unique competitive advantage in the digital landscape.