Xiaohongshu, the lifestyle and social commerce platform known in the West as RedNote, has made its most significant organizational commitment to artificial intelligence, announcing the creation of “Dots,” a new first-level AI department that consolidates the company’s scattered AI initiatives under a unified leadership structure. The move signals a decisive break from the platform’s historically cautious approach to AI integration.
From Hi Lab to Dots: An Organizational Upgrade
On April 30, Xiaohongshu sent an internal letter to all employees announcing the formation of Dots, which was upgraded from the former Humane Intelligence Lab (Hi Lab). The new department encompasses four divisions: model R&D, infrastructure, engineering, and product. It reports directly to Ding Ling, known internally as Conan, who was recently appointed as Xiaohongshu’s president. Ding Ling brings a background that spans computer engineering (Nanyang Technological University), an MBA from Stanford Graduate School of Business (class of 2011), and stints at BCG and Citi.
The appointment is notable because Ding Ling and Xiaohongshu founder Mao Wenchao (alias: Seiya) were both at Stanford GSB in 2011, giving the new AI leadership structure a foundation of long-standing personal trust. Xiaohongshu’s chief product officer, Deng Chao (alias: Yingmu), who joined as the company’s third employee and has an architecture background, rounds out the senior team overseeing the platform’s transformation.
Diandian’s Struggles Reveal the Challenge
The creation of Dots comes amid mixed results for Xiaohongshu’s existing AI product. Diandian, the company’s standalone AI application, launched in August 2024 with an in-house foundation model. After that model underperformed, Diandian was connected to the DeepSeek LLM on February 26, 2025. Despite multiple iterations, including a v2.0 release in July 2025 and a strategic refocus on community features in November 2025, the app has struggled to gain traction. A late-April review by 36Kr found Diandian ranked only 186th on app store download charts, a stark contrast to ByteDance’s Doubao, which had accumulated 1.92 million ratings compared with Diandian’s 45.
Xiaohongshu also operates Wenyiwen, an AI-powered search product embedded within its community platform, which uses the open-source Qwen model. The fragmentation of these efforts across different teams and mandates is precisely what the creation of Dots is designed to address. By consolidating model R&D, infrastructure, engineering, and product under a single department, Xiaohongshu is attempting to create the organizational coherence that effective AI product development requires.
Product Innovation Over Foundational Research
Xiaohongshu’s AI strategy, as articulated through the Dots structure, prioritizes product-level innovation rather than foundational model development. The company is not attempting to compete with DeepSeek or Moonshot AI on benchmark performance; instead, it is focused on integrating AI capabilities into the specific context of its community, a platform built around user-generated content, lifestyle discovery, and social commerce. The challenge is to deploy AI in ways that enhance the authenticity and serendipity that make Xiaohongshu distinctive, rather than reducing it to another generic chatbot interface.
This approach aligns with broader industry trends, where platforms are grappling with how to deploy AI in ways that enhance rather than disrupt the user experience, a challenge that intersects with the ongoing debate over AI agent regulation and autonomous governance. The creation of Dots suggests that Xiaohongshu has concluded that the window for cautious experimentation has closed and that the platform must now move with urgency to remain relevant in an era where AI is reshaping how users discover, create, and share content.
The Stakes: Xiaohongshu’s Unique Position
Xiaohongshu occupies a distinctive position in China’s social media landscape. Unlike WeChat, which is a utility platform, or Douyin, which is optimized for passive video consumption, Xiaohongshu is built around active discovery, users come to the platform to research products, plan travel, find recipes, and explore lifestyle ideas. This intent-driven usage pattern makes the platform particularly well-suited for AI augmentation: an AI assistant that understands a user’s interests, surfaces relevant content, answers questions, and facilitates transactions has a natural home in a discovery-oriented community.
The challenge is that Xiaohongshu’s community culture is deeply rooted in authentic, user-generated content. Aggressive AI integration risks undermining the trust and authenticity that differentiate the platform from its competitors. The Dots department’s mandate to integrate AI at the product level rather than the model level reflects an awareness of this tension and a commitment to finding an approach that enhances the platform’s distinctive qualities rather than homogenizing it into another generic AI interface.
Whether Dots can deliver on that mandate will depend on execution. The organizational upgrade is necessary but not sufficient: Xiaohongshu will need to ship products that users actually want, at a pace that keeps it competitive with ByteDance’s Doubao and the broader wave of AI assistants that are reshaping how Chinese consumers interact with digital platforms. The next twelve months will be a critical test of whether the platform’s belated urgency translates into products that matter.