AI News

Tencent launches AI interactive knowledge product "Dongle Ya", why are big companies rushing to bet on AI education

36Kr · AIUpdated Nov 16 19:17

AI summary

Tencent PCG launches AI interactive knowledge product "Dongle Ya"; users can use AI to turn knowledge into dynamic courseware, 3D demonstrations, interactive picture books and other content, and it supports sharing and secondary creation, with the goal of building an "AI version of electronic Lego" for teenagers. The product was officially approved in September this year and is currently operated as an independent project; "Qimiao Community" already includes more than ten content categories, but the external creator incentive mechanism and commercialization path are not yet clear. Since 2026, domestic big companies such as ByteDance and Alibaba, as well as OpenAI, Google, and Anthropic, have also been intensively expanding AI learning scenarios.

Body · AI translation

Recently, the AI interactive knowledge product "Dongle Ya" created by Tencent PCG (Platform and Content Group) has attracted attention. Users can use AI to transform knowledge into dynamic courseware, 3D demonstrations, interactive picture books and other content, and it supports sharing and secondary creation. Tencent hopes to use this to create an "AI version of electronic Lego" for teenagers.

Dongle Ya began exploration at the end of 2025, was officially established as a project in September this year, and is currently operated as an independent project. This is also a new attempt by Tencent in the field of interactive knowledge content.

Beyond Tencent, since 2026, domestic internet giants such as ByteDance and Alibaba have continued to launch new AI education features, and overseas tech giants are also accelerating the expansion of learning scenarios. From photo-based question answering to AI lectures and interactive content generation, the education layouts of major companies are extending to more learning stages.

The birth of Dongle Ya stems from an observation by the Tencent team of museums, exhibitions and other scenarios.

In museums, complex knowledge is often presented through models, multimedia displays and interactive experiences, and visitors can operate while viewing and gradually understand the principles involved. The Tencent team thus came up with an idea: can AI enable ordinary people to easily create such content?

At the end of 2025, the team turned this idea into a Vibe Coding project called Eureka. Users input a word, and the system can generate multimedia content similar to a mind map. The project once ranked third on Product Hunt's global daily list. Since then, the team has continued to explore AI creation tools for ordinary users and officially established Dongle Ya as a project in September this year.

In practical applications, Dongle Ya can transform text and knowledge points into dynamic courseware, 3D demonstrations, interactive picture books and mini-games. Taking subjects such as biology and physics as examples, in the past teachers often had to make experiment demonstrations or courseware themselves, and students mainly understood knowledge by watching. With Dongle Ya, teachers can turn complex knowledge points into operable interactive content, allowing students to observe changes under different conditions by dragging, clicking and other methods.

Such applications can also extend to family scenarios. When children become curious about extracurricular knowledge such as rocket launches and natural phenomena, parents can look for existing works or use AI to generate relevant demonstrations. Even without professional capabilities such as programming and animation production, ordinary users can participate in the creation of knowledge content.

In Dongle Ya, content creation and the community are placed together. At present, its "Wonderful Community" already includes more than ten content categories, covering subjects such as Chinese, mathematics and English, as well as fields such as nature, culture, health and science. Users can directly browse and use existing works, and can also modify, improve and continue creating based on the original content.

Behind this design is Tencent's exploration of the way knowledge content is supplied. In the past, a set of interactive courseware was usually produced by professionals, and users were mainly responsible for viewing and operating it. Today, AI has lowered the production threshold, and teachers, parents, education and training practitioners and ordinary people with professional knowledge all have the opportunity to become content producers. The same knowledge point can also derive multiple forms of expression according to different ages, interests and learning needs.

But after the amount of content increases, how to ensure that knowledge is accurate and interaction is effective has also become a problem that the platform must face.

At present, Dongle Ya tries to call knowledge bases or material libraries during the generation process, and controls quality by combining curriculum standards, verifiable materials and manual verification. After a work is generated, the system will also verify it by simulating clicks and checking animation frames. As the scale of content expands, the team plans to further introduce user cross-review and content improvement mechanisms.

At this stage, the content of Dongle Ya is still mainly driven by the team, the incentive mechanism for external creators is still being explored, and the commercialization path is not yet clear. The team currently focuses mainly on content quantity, quality and user retention, hoping to first establish a batch of high-quality examples and then gradually attract more creators to participate.

From explaining questions to giving classes, domestic giants compete for the A1 learning entrance

Since 2026, internet companies such as ByteDance and Alibaba have also been intensively updating learning products, and the role of AI is gradually extending from answering questions to knowledge explanation, course generation and interactive teaching.

At the end of August, Tencent Yuanbao launched "AI Jingjiang", combining dynamic blackboard writing with two-way voice interaction. After students take photos and upload questions, AI will break down the problem-solving ideas, explain step by step with real-time blackboard writing, and check students' understanding through interactive questions. During the explanation, students can also interrupt and ask follow-up questions at any time.

This also allows Yuanbao to further complete its teaching and explanation capabilities beyond existing functions such as photo-based question answering and homework correction. In June this year, Tencent also partnered with the QQ browser to launch "Yuanbao Gaokao Tong", extending AI services to gaokao consultation and volunteer application.

In May, Doubao Aixue launched "Doubao Classroom", applying ByteDance's video generation model Seedance to course production. Taking "Seven-Step Poem" as an example, AI first restores the poem's background through a historical scene short film, then explains knowledge points with blackboard writing, pictures and interactive questions, forming a course of about 10 minutes. In addition to premium courses produced by the team, users can also customize exclusive courses according to their own learning needs.

ByteDance's overseas product Gauth further attempts to change the way courses are presented. After launching the visual knowledge landscape Atlas in May and AI Course in August, Gauth recently launched the "Infinite Digital Blackboard" and upgraded AI Course to Gauth Academy.

On this blackboard, the AI tutor can generate notes, charts and 3D interactive content while explaining. Students can freely view the connections between different knowledge points and ask questions at any time, letting AI supplement explanations directly on the canvas. Compared with video courses played in a fixed order, this method gives students more room for active exploration.

ByteDance uses video generation and multimodal interaction capabilities for courses, while Alibaba continues to improve functions around high-frequency learning needs such as textbooks, homework and wrong questions.

3 In [month], the Qwen App launched "AI Teaching Materials (Mini Lecture Hall)", while also upgrading photo-based question answering, and rolling out features such as full-page homework grading, automatic organization of a wrong-answer notebook, and re-practice of wrong answers. Students can enter learning through textbook explanations, then discover problems through homework grading, and use wrong-answer practice to consolidate knowledge. In July, Qwen also launched a free summer AI general education course for elementary school students, extending AI applications to dialogue, exploration, and creative practice.

Behind these updates is a common product approach: letting AI take on more teaching steps and keeping the process from students asking questions to understanding knowledge and completing practice on the same platform. As basic question-answering capabilities gradually become widespread, course content, interactive experience, and continuous learning services are becoming new priorities for major tech companies competing for education users.

From personal learning to enterprise training, overseas tech giants broaden their education footprints

First is OpenAI, with a focus on improving ChatGPT's personal learning experience.

In March this year, OpenAI launched dynamic interactive explanation features for math and science in ChatGPT, initially covering more than 70 core concepts. Students can directly adjust variables in formulas, observe how charts and calculation results change, and understand abstract knowledge through hands-on operation.

Before that, OpenAI had launched Study Mode, which supports learning through step-by-step guidance, interactive questioning, and knowledge checks. This update further strengthens ChatGPT's visual explanation capabilities. OpenAI hopes students can not only get answers through ChatGPT, but also understand knowledge and explore problems within the same conversation.

Google, by contrast, focuses more on connecting AI learning capabilities with existing education products and further entering school teaching scenarios.

In June, Google launched the Study Notebooks feature in Gemini. After students upload materials such as textbooks and class notes, the AI identifies knowledge weak points through diagnostic quizzes, generates targeted learning content, and adjusts the learning plan based on subsequent quiz results.

Relying on Google Classroom, Google is also advancing teacher-led AI learning activities and plans to support teachers in assigning personalized learning tasks based on course materials and viewing students' learning progress. Compared with standalone learning apps, Google can use existing school teaching platforms to connect AI with lesson preparation, course learning, and teaching management.

Anthropic, meanwhile, has further extended its education business into enterprise talent development.

On October 2, Anthropic announced a $100 million investment to launch Claude Frontier Academy, planning to train 10,000 forward-deployed engineers by the end of 2027. The first batch of participating companies includes Accenture, Bain, Deloitte, McKinsey, and Morgan Stanley, among others.

The project conducts training directly around enterprises' AI deployment needs. Trainees need to complete simulated enterprise deployment tasks and practical assessments, and after passing, enter a 12-week practice phase to advance real AI projects at their companies with support from Anthropic engineers.

Previously, Anthropic had launched Claude for Education for universities and Claude for Teachers for primary and secondary school teachers. This further entry into enterprise training connects AI skills development with real enterprise applications and also provides talent support for promoting its enterprise-level products.

The continued entry of major tech companies is changing the competition rules in the education industry.

As capabilities such as AI problem explanation and course generation gradually become basic features of general-purpose products, the commercial value of single tools and standardized content will come under pressure, and education companies will also face new customer acquisition and pricing pressure. Competition is further shifting to teaching service steps. For tutoring institutions and education technology companies, professional teaching research, learning diagnosis, personalized teaching, and long-term effect tracking will become more important competitive barriers.

In the future, major tech companies may occupy the learning entry point through technology and traffic, while professional education companies need to prove their value in teaching effectiveness and service delivery. The division of labor in the education industry may thus be reshaped: who controls the learning entry point, who provides teaching services, and who can truly be responsible for learning outcomes?

This article comes from the WeChat public account "Duojing" (ID: DJEDUINNO), author: Siluo, published by 36Kr with authorization.

Source

36Kr · AI