Building AI-Native
Learning Systems
I founded COCOROBO in 2016 with a simple conviction: technology should not just deliver content — it should reshape how learning happens. Over the past decade, we have built and deployed AI learning products and learning systems in 1,200+ schools, working alongside thousands of teachers and hundreds of thousands of students.
This site is a window into that journey — the products we build, the learning sciences that inspire us, the thinking that grounds our work, the research that emerges from real classrooms, and our vision for AI-native education — not AI tools for education, but systems where learning becomes continuous, adaptive, and collaborative.
COCOROBO Product Ecosystem
View allCocoClass
Classroom orchestration — making every lesson interactive and observable
Haiyang (Tony) Xin 辛海洋
Founder & CEO, COCOROBO
Tony leads COCOROBO, serving 400+ schools in Hong Kong & Macau and 800+ schools across the Greater Bay Area, building learning systems in real classrooms for over a decade.
With a Ph.D. from CUHK and experience as an MIT visiting scholar, he connects learning sciences, AI system design, and educational practice.
Full bio1,200+
Schools Served
10+
Years in EdTech
13
Publications (2024–26)
5
Products in System
Learning systems, not tools
We design learning systems where AI is part of the structure of teaching and learning — spanning classroom orchestration, collaborative knowledge building, personalized learning, and AI agent creation.
From STEM tools to AI-native systems
Since 2016, COCOROBO has designed and developed a complete product ecosystem. We started with STEM education hardware (CocoMod) and AI education tools (CocoPi), then evolved into learning platforms (CocoClass, CocoNote, CocoStudy), and now into Agentic AI infrastructure (CocoFlow). Along the way, we have served over 1,200 schools, hundreds of thousands of students, and thousands of teachers.
From classrooms to academic contribution
Our approach begins with frontline practice: we discover problems in real classrooms, analyze them with rigorous academic methods, and solve them through technology and system design. Since 2024, we have published continuously at AERA, ISLS (CSCL, ICLS), and AIED.
Why it matters
Most AI in education focuses on efficiency. But the more fundamental question is:
How does learning change when AI becomes part of the system?
This is about rethinking how learning processes are organized — from isolated tool use to system-level intelligence, from individual interaction to coordinated orchestration, from surface-level automation to deep infrastructure.
We believe the next stage is AI-native learning systems — where classrooms function as adaptive ecosystems and AI participates in the constitution of learning itself.