Junyi Academy’s Five Education AIs, Remembered in Four Lines
Illustration · not a service recordJunyi Academy’s five AI learning products are here
The recent PISA 2025 reminds us that the relationship between AI use and learning performance varies with the purpose, the frequency, and whether schools provide related learning opportunities.
The OECD’s recent synthesis of a number of studies also points out that only AI with a clear pedagogical purpose and guidance stands a better chance of turning “completing the task” into “gaining the ability.”
So we need to move the discussion of “whether or not to use AI” forward, to:
Which AI? How should it be used? What should children learn from it?
This is the core of Pedagogical AI, or education AI: building “how children come to learn” into the design of the AI. It is not only bringing AI into education, but bringing education into AI.
When practice is needed, there is practice; when a child is stuck, there is a hint; after feedback, there is a chance to try again.
And learning languages, exploring society, getting to know AI, and understanding math and science each call for a different kind of practice. So Junyi has designed a series of education AIs, which can be remembered in four lines:
① Learn to read
Jutor, the English AI tutor, and Lingoleap, the Chinese language arts AI tutor (Beta)
From listening and speaking to reading and writing, they give children more chances to practice language and get feedback, and more courage to speak up, read with understanding, and express themselves.
② Read to learn
PlayGo AI
It starts from a single “why,” whether in social studies, across subjects, or in everyday life, keeps the questions coming through conversation, and turns curiosity into the starting point for exploring knowledge.
③ Learn to use AI
ActiveAI
Practice “Ask, Use, Manage, Build (問用管造)” together: ask clearly, use appropriately, know how to verify and take responsibility, and be able to create things with your own hands.
④ Use AI to learn
Junyi AI Notebook
When learning math or science, if a video leaves a question or a problem has a child stuck, the child raises the part they don’t understand and clears it up step by step in conversation with the “semi-Socratic AI teaching assistant.”
Teachers can also use ShareClass to find teaching resources, prepare lessons, and share their own good methods.
Today, we invite teachers, parents and students to bring a question they have not yet figured out and give Junyi a try.
It can be a passage of English, a question from a textbook text, or a math problem. After practicing, try saying it in your own words: “What do I understand now?”
We hope that every child who has ever said “I can’t do it” will have more chances to say:
“So that’s how it is. I’ve got it!”
Start here: Junyi Academy and beyond
- Original: Ray Lu’s personal Facebook post (published 2026-09-09)
Course titles follow the source. Check each course page for its teaching language.
Dissecting the Crawfish: How AI Agents Work, via OpenClaw
An OpenClaw-based introduction to how AI agents work. Free public lecture; no self-study certificate. Costs of independently using external models or tools are separate.
Introduction to Model Context Protocol
Build MCP servers and clients with Python tools, resources, and prompts. Python/async familiarity required; connected model API usage may incur charges.