Our Mission
A shared layer for human knowledge
For most of history, good teaching has been a privilege of where you were born, what your family could afford, and which schools were within reach. A child within walking distance of a great library, a great teacher, or a great university has gates open to her that a child without those resources, twenty miles or twenty thousand miles away, may never see. The internet narrowed this gap by making information available. It did not close it. Information is not knowledge. A textbook PDF is not a teacher. A video lecture is not a teacher.
We are building the missing layer: a shared hub where the world’s knowledge is not just available, but taught.
What this looks like
The kind of moments we’re building for
A teacher in rural Kenya
She opens xLearnHub and finds the same physics lessons that students at MIT study from, adapted in seconds to her syllabus and her language. Her students don’t just read the lessons. They study them with an AI tutor that answers their questions at midnight, in their own language, with the patience their teacher cannot give to forty children at once.
A high school student in Lagos
She dreams of becoming a doctor. She sits down to study biology. The course she opens was originally written by a Stanford professor, forked and improved by a Lagos teacher who knew which examples would resonate, and tutored by an AI that knows the entire content in depth. She’s no longer alone with her textbook. She’s no longer dependent on whether her school can afford to hire enough teachers. The world’s best teaching is hers, for free.
A grandfather in Mumbai
He never finished school. He decides to learn algebra alongside his granddaughter. The patient tutor that walks him through the basics doesn’t make him feel embarrassed for asking the same question three times. He finishes the chapter. He starts the next one. He keeps going.
This is what becomes possible when knowledge is treated as shared infrastructure, not a product locked behind a paywall, not a static file on a forgotten server, not a video that played once for a million people, but a living, growing, taught body of human understanding.
Why now
Three shifts came together in the last two years.
What makes those moments possible now, and not five years ago, is a specific convergence in AI. For xLearnHub to work, three things had to line up at the same time. They finally have.
Shift 1
AI coding tools matured.
Cursor, Claude Code, Codex changed how software gets built. The vibe-coding paradigm became standard: you direct AI in plain English, and AI builds. We’re applying the same paradigm to knowledge. Where developers got AI that writes the code, creators get AI that builds the interactive lesson.
Shift 2
Frontier model quality crossed the teaching threshold.
Modern models, grounded carefully in source content, teach better than the median TA in many subjects. AI tutoring stopped being speculative and became reliable. Cheaper inference made per-learner memory architecturally feasible. Agents can run continuously between sessions, not just during them.
Shift 3
Open educational resources are finally large enough to seed a network.
Twenty years of OER (OpenStax, Oak National Academy, MIT OCW, Khan Academy, OER Commons) have produced an enormous body of openly-licensed material. Until now it sat as static files. We can ingest it, make it interactive, AI-tutor it, and make it forkable on day one. The cold-start problem that killed previous remix platforms is mostly solved before we begin.
Three converging shifts. The primitives are proven. The moment for this is now.
From the founder
Why I’m building this
I’m Lingzhi Kong, a machine learning engineer who grew up in a small village near the northern border of Inner Mongolia, China, an underdeveloped region where educational resources were in great shortage and most students never made it to college. I taught myself out of it, improving my college entrance exam score by nearly 300 points in twelve months, alone with a textbook, no tutors, no one to ask when I was stuck.
That experience is why I built this. The pain points xLearnHub solves: being stuck at midnight with no feedback, not knowing if your understanding is right or quietly broken, having no idea what to study next. I didn’t learn them from a market study. I lived them.
I see the shift every day in my own work. I build xLearnHub using Claude Code and Codex, the same vibe-coding paradigm I’m bringing to creators. The product I’m shipping is the tool I use, one level of abstraction up. That’s why I see the analogy others miss.
In software & coding
In learning & teaching
Software engineer
builds software
Instructor / creator
teacher, professor, YouTuber, author
GitHub
shared, forkable, improvable code corpus
xLearnHub
shared, forkable, improvable knowledge corpus
Vibe coding
direct AI in plain English, AI builds the code
Vibe authoring
direct AI in plain English, AI builds the lesson
GitHub Copilot
pair programmer, always on
AI tutor
pair learner, on every page, in the creator’s voice
Why me. AI coding tools changed how software gets built. I’m building xLearnHub to do the same for instructors, creators, and learners. The technical revolution that already happened for software is overdue for education, and the right person to bring it is someone who has lived the educational gap and built modern AI infrastructure on the other side. That’s me.
How we built it
Three foundations, AI-native from day one
AI-native is an architectural commitment, not a feature. The content data model is designed for AI agents from the ground up. The authoring agent, the in-canvas vibe builder, the per-learner background memory agent, the tutor on every page: every system is AI-native, and AI-centered from day one. All of it built for humans, for creators and learners. Three foundations follow.
Vibe creation
Anyone can turn what they already have (a lecture recording, a PDF, a set of notes, a YouTube video, or just an idea) into a polished interactive lesson by typing in plain English. No course-platform configuration, no LMS, no web development. The AI does the technical work; the creator decides what gets taught and how. The same paradigm that AI coding tools brought to writing software, applied to authoring online learning.
AI tutoring, customized three ways
Every published lesson ships with a tutor that is grounded in the content (knows the specific lesson in depth), shaped by the creator’s voice (tone, rigor, the analogies the creator uses), and personalized to each learner (between sessions, a background memory agent updates what the tutor knows about them: practice history, strengths and stuck-points, what to suggest next). Not a generic chatbot. The missing one-to-one relationship that broadcast content destroyed, restored at the scale only AI makes possible.
Forkable knowledge
Every lesson can be marked remixable. Other creators can fork it, adapt it, extend it, and republish under their own name with attribution preserved. Quality compounds the way it does on GitHub: the best explanations get forked the most, the best forks get studied the most, and the people who do the work earn for it.
None of this could be bolted onto a pre-AI platform. The data model is the difference. AI-native infrastructure. Human creators and learners.
What this becomes
A hub for human knowledge. For everyone.
When a great teacher publishes once on xLearnHub, her teaching becomes a living mentor that can be in a million places at once. The kid in rural Kenya studies with the same chemistry teacher as the kid at Cambridge. The high schooler in Lagos gets the same MCAT tutor as the student at Stanford. The grandfather in Mumbai learns from the same patient algebra teacher his granddaughter does.
Geography stops being destiny. Expertise stops being scarce. The teachers who shape our minds finally reach everyone they could help, at midnight, in any language, in every corner of the world that has been waiting.