Xiao Yuan Learning Machine Launches the “Primary and Secondary School Textbook Learning Agent”
On August 21, 2026, with the support of the People’s Education Audio-Visual Digital Publishing House, Xiao Yuan Learning Machine held the launch event for the “Primary and Secondary School Textbook Learning Agent” in Beijing. Led by the People’s Education Audio-Visual Digital Publishing House, the agent was jointly developed on the foundation of high-quality textbook content resources and Xiao Yuan’s self-developed Yuanli large language model. Through a dual-mode learning system combining “AI Learning Companion + AI Learning Guide,” it provides students with personalized AI solutions for textbook learning.
On the day of the event, the two parties formally signed a cooperation agreement and announced the official launch of the “Primary and Secondary School Textbook Learning Agent” on Xiao Yuan AI Learning Machines. Xiao Yuan Learning Machine has become the first smart hardware channel to distribute the “Primary and Secondary School Textbook Learning Agent,” bringing authoritative textbook content directly into students’ self-directed learning scenarios in the form of an AI learning assistant.
Breaking Through in Intelligent Learning Support: Authoritative Content and Educational AI Join Forces
In April 2026, the Ministry of Education and four other ministries jointly issued the Action Plan for “Artificial Intelligence + Education”, explicitly proposing to “leverage artificial intelligence to empower student learning. With a focus on promoting the all-round development of students in morality, intelligence, physical fitness, aesthetics, and labor, develop intelligent learning companions,” and to “integrate educational foundation models and intelligent agent tools to create a range of thematic learning scenarios.” Driven by these policies, developing intelligent agent products centered on authoritative textbooks and designed to systematically serve learning scenarios has become a critical missing piece that the education industry urgently needs to address.
How can this goal be achieved? In his speech, Long Zhengwu, General Manager and Editor-in-Chief of the People’s Education Audio-Visual Digital Publishing House, provided the answer: “Only when content and technology come together can we create truly effective educational products.” By planning and developing the “Primary and Secondary School Textbook Learning Agent,” the People’s Education Audio-Visual Digital Publishing House hopes to deeply integrate authoritative textbook content resources with AI capabilities, serve students’ self-directed learning, and address challenges such as students being unable to understand what they read, lacking depth in their learning, and having no access to real-time interaction.
The technological foundation for addressing these challenges lies in Xiao Yuan’s years of accumulated strengths in educational research, data, products, and AI capabilities. Trained jointly on years of educational research and user data, Xiao Yuan’s educational AI has a deep understanding of textbooks, students, and the learning process. It not only knows “how to solve a problem,” but also understands “why a child cannot solve it, where they are stuck, and what they should learn next.”
Xiao Yuan’s software and hardware products have served more than 500 million users in total. Through its efforts to transform AI capabilities into product experiences that users can perceive and rely on, Xiao Yuan has accumulated extensive practical experience, ensuring that its intelligent agent products can operate reliably and continuously in real-world, complex, and high-frequency learning scenarios. Content safety, technological maturity, and system reliability provide essential safeguards for turning the “Primary and Secondary School Textbook Learning Agent” from a conceptual vision into a mature product.
“The integration of artificial intelligence and educational publishing is neither simple content processing nor the addition of features. It is about enabling high-quality content to genuinely become part of the learning process while following the principles of publishing and education,” said Wang Xiangdong, Vice President of Yuanli Technology Group and CEO of Xiao Yuan Learning Machine. High-quality textbook content answers the question of “what to learn,” while educational AI capabilities address “how to learn it.” The deep integration of quality educational publishing content with advanced artificial intelligence capabilities further incorporates textbooks into children’s learning processes—moving from “being accessible” to “being truly understood,” and from “digital resources” to “intelligent learning services.”
At the event, witnessed by Long Zhengwu, General Manager and Editor-in-Chief of the People’s Education Audio-Visual Digital Publishing House, and Wang Xiangdong, Vice President of Yuanli Technology Group and CEO of Xiao Yuan Learning Machine, the cooperation agreement was signed by Wei Xiaoxi, Deputy General Manager of the People’s Education Audio-Visual Digital Publishing House, and Zhang Lanpu, Deputy General Manager of Marketing at the Tools Business Unit of Yuanli Technology Group, on behalf of both parties. In the future, the two parties will continue to deepen their cooperation and promote the large-scale adoption of the “Primary and Secondary School Textbook Learning Agent” through the smart hardware ecosystem.
From “Having Studied” to “Having Learned”: The Dual-Mode System Redefines Textbook Learning
For most children, textbook learning at home has long faced structural challenges: they do not know where to begin, cannot tell which content they have failed to master after studying, have no one to ask when they encounter questions, and lack both a clear path to follow and a guide who is always available.
Unlike the “tool-oriented” educational AI products available on the market, the “Primary and Secondary School Textbook Learning Agent” is built around a dual-mode system of “AI Learning Companion + AI Learning Guide,” providing a systematic learning solution. The AI Learning Companion addresses “whether students are willing to learn,” while the AI Learning Guide addresses “whether they know how to learn.” The two modes provide support for children with different levels of self-directed learning ability and complement each other across different learning scenarios.
In “AI Learning Companion” mode, static textbook images and text are transformed into dynamic AI textbooks, with lesson content presented through visual scenario-based animations. Static textbooks are brought to life. On this foundation, features such as tap-to-read and read-along practice, recitation assessment, character and vocabulary learning, and AI Q&A are introduced in sequence, giving children a learning experience in which “someone explains, someone asks questions, and someone accompanies them.”
The “AI Learning Guide” mode addresses the depth of learning. Through one-on-one interactive instruction, the AI teacher relies on curriculum standards and the logic used to compile the textbooks to intelligently plan learning paths by educational stage, subject, and lesson type. It also calls on high-quality digital textbook content resources—including text, images, audio, video, and animation—in real time to support children’s learning.
Take primary school Chinese as an example. The AI teacher guides students through four steps: “Read the text → Learn the characters and words → Understand the general meaning of the text → Think and appreciate.” Guided questions encourage children to think actively and systematically master the textbook content. This learning process was not designed in isolation; it is the result of in-depth collaboration between the educational research teams of both parties. The teaching objectives, lesson plans, identification of key and difficult points, and depth of AI intervention for each class were determined through lesson-by-lesson discussions between professional editorial and educational research teams.
Content safety runs throughout the entire product development process. During development, the teams consistently implemented strict content review procedures and integrated safety capabilities into the entire process of model training, understanding, and generation through an AI-native security system. At the same time, they established an end-to-end mechanism covering input recognition, generation control, output review, and risk feedback. The two parties jointly built clear data-calling and content-safety mechanisms from the ground up, ensuring that every stage has standards, boundaries, and safeguards.
With the coordinated operation of the “AI Learning Companion + AI Learning Guide” dual-mode system, textbook learning has undergone a structural transformation: from “finishing a lesson” to “truly understanding it,” and from static, one-size-fits-all presentation to active exploration tailored to every learner.
Knowledge-Point Mastery Increased by 33.6%—Efficient Learning Becomes a Reality
In March 2026, the “Primary and Secondary School Textbook Learning Agent” was piloted on the first-generation Xiao Yuan AI Learning Machine models, attracting more than 100,000 primary and secondary school students to date.
Pilot data shows that among students who used the “Primary and Secondary School Textbook Learning Agent” consistently, mastery of Chinese-language knowledge points increased by 33.6%, while mastery of English-language knowledge points increased by 24%. This indicates that learning did not remain at the superficial level of merely “finishing the textbook,” but genuinely completed the cognitive transformation from input to understanding. Students’ speaking rates in Chinese and English reached 81.4% and 61.7%, respectively. Guided by the intelligent agent, students thought and expressed themselves more actively, with significant increases in engagement and autonomy.
At the same time, parents’ average daily time spent accompanying their children’s studies decreased by 20 minutes. “Less stressful, effective, safe, and trustworthy” became the three most frequently mentioned keywords in parents’ feedback. One parent user from Hangzhou, Zhejiang, said: “In the past, helping my child preview Chinese meant reading the text once and going over the characters and words. Now my child learns independently with the intelligent agent—from reading the text and learning the characters and words to understanding the text, every step is guided. My child is more willing to speak and is genuinely learning.”
The launch of the “Primary and Secondary School Textbook Learning Agent” on Xiao Yuan AI Learning Machines marks not only an early exploration by smart terminals represented by Xiao Yuan Learning Machine, but also provides a practical example of the implementation of digitalization in basic education. As textbooks evolve from “static text” into dynamic learning systems that are multimodal, intelligent, and highly interactive, the way students learn from them is being redefined. As Wang Xiangdong put it: “When educational AI returns to textbooks and to the starting point of children’s self-directed learning, high-quality learning companionship will become a normal experience available to every child.”