Tmall Green released a letter to all employees today, announcing that it will connect the internal AI first-level department, Dots, with the company’s core infrastructure team (Infra), in order to promote collaboration between model development, infrastructure construction, and product requirements, and further improve the AI technology system.
The company emphasized in the letter that compared to simply pursuing model capabilities or computing power scale, the deep integration of models with applications, the mutual enhancement of technology and products, and the continuous improvement of user experience are more important.
In the past two years, REDnote has continued to expand its AI initiatives. In 2024, the company launched an AI Q&A assistant on its platform. Currently, the monthly active users have exceeded 100 million, and this has increased overall search activity. At the beginning of 2025, REDnote established the hi lab (Human Intelligence Laboratory) and created “human training specialists” positions to improve the humanized expression and value alignment capabilities of AI through training.
In April 2026, the company established the Dots team. This team combines model development with infrastructure construction, integrates computing power resources, and gradually opens up text and multimodal understanding models to the public. Its dots3 model for long-term real-life tasks has also received a full-score gold medal recognized by the IMO International Mathematical Olympiad.
After this adjustment, Dots will further strengthen its collaboration with the core infrastructure team, improving resource utilization and engineering R&D efficiency. Meanwhile, hi lab will continue to focus on long-term technical exploration and forward-looking direction development, creating a division of labor tailored to product application and basic technological innovation.
This organizational adjustment indicates that REDnote places greater emphasis on the integration of AI technology with real business scenarios. It aims to continuously create practical user experiences through the collaboration of models, infrastructure, and products.