Xiaoyuan Learning Machine Launches AI Textbook Learning Agent with People's Education Press

During the event, the two parties signed a cooperation agreement and announced that the agent would be available on the Xiaoyuan AI Learning Machine, making it the first smart hardware device to offer this textbook-focused AI agent. The agent, developed by People's Education Audio-Visual Digital Press, leverages high-quality textbook resources and Xiaoyuan's self-developed Yuanli large model to provide personalized textbook learning solutions.

The launch follows the release in April 2026 of an action plan by the Ministry of Education and four other government departments, which called for using artificial intelligence to empower student learning and for developing AI learning companions and intelligent tools. Long Zhengwu, General Manager and Editor-in-Chief of People's Education Audio-Visual Digital Press, said that "content and technology must work together to create a good educational product," and that the agent aims to solve students' problems of not understanding textbooks deeply enough and lacking real-time interaction.

Xiaoyuan's AI capabilities are built on years of educational research and user data. Its educational AI understands not only how to solve problems but also why a child struggles and what to study next. Xiaoyuan's software and hardware products have served more than 500 million users, providing a reliable foundation for deploying the agent in real, complex learning scenarios.

Wang Xiangdong, Vice President of Yuanli Technology Group and CEO of Xiaoyuan Learning Machine, said that the combination of AI and educational publishing is not simple content processing or feature stacking, but a way to integrate quality content into the learning process. High-quality textbooks provide what to learn, while educational AI helps students truly master it.

The agent's core architecture is a dual-mode system. In "AI companion" mode, static textbook content becomes animated and interactive, with features such as read-aloud, recitation assessment, vocabulary learning, and AI Q&A. In "AI guided learning" mode, an AI teacher provides one-on-one interactive instruction, planning learning paths according to curriculum standards and textbook logic, and dynamically calling on digital resources such as text, images, audio, video, and animation.

For example, in primary school Chinese language, the AI teacher guides students through four steps: reading the text, learning words, understanding the main idea, and thinking and appreciation. This process was co-designed by editorial and teaching research teams, with each lesson's goals, syllabus, and AI involvement depth reviewed item by item.

Content safety was a priority throughout development. The team maintained strict content review, integrated an AI-native security system into training and generation, and built a full-chain mechanism covering input recognition, generation control, result review, and risk feedback.

Pilot deployment began in March 2026 on the Xiaoyuan AI Learning Machine, attracting over 100,000 primary and secondary school students. Pilot data showed that students who consistently used the agent improved their knowledge mastery in Chinese by 33.6% and in English by 24%. Speaking rates in Chinese and English reached 81.4% and 61.7% respectively. Parents reduced daily accompanied study time by 20 minutes, and three most frequent feedback words were "saving worry, effective, and trustworthy." A parent from Hangzhou said that previously previewing Chinese meant reading aloud and recognizing words, but now the child follows the agent with guidance and is willing to speak and truly learns.

The debut on the Xiaoyuan learning device represents a practical sample for the digitalization of basic education. As Wang Xiangdong stated, "Educational AI returns to textbooks and to students' independent learning; high-quality learning companionship will become a normal state that every child can enjoy."