CocoRobo Research logo
CocoRobo · Research & Practice

CocoRobo Research is CocoRobo’s research and practice platform for building AI-native learning infrastructure with schools, teachers, students, and long-term university partners.

We treat every product as a research platform, observing how learning actually unfolds in real classrooms and iterating through design-based research — so that learning-sciences theory, agentic AI architecture, and frontline teaching practice continually inform and advance one another, accumulating into learning infrastructure that can be studied and reused.

Scale in schools

1,400+schools served across Hong Kong, Macau, and the Greater Bay Area
5,000+teachers reached through products, training, and school programmes
100,000+students impacted through AI-native learning platforms and classroom practice

Research & proposal work

DLW 2026proposal supporting pages prepared for submission
11selected research outputs across AERA, ISLS, AIED, and CSCL
5long-term university collaborators
AI-native learning systems map Learning, teaching, inquiry, collaboration, evidence and practice are connected by orbit-like system lines around AI-native learning systems. Operational entities (agents, traces, pilots, data) feed the system. Learning Teaching Evidence Inquiry Collaboration Practice AI-Native Learning Systems Agents Traces Pilots Data

AI-Native Learning Systems

Reorganizing learning, teaching, collaboration, evidence and educational practice.

CocoRobo Research
Research Manifesto

From supplementary AI use to AI-native learning systems.

CocoRobo Research keeps the practical classroom problem in view while asking a deeper research question: how does AI change the organization of learning, teaching, collaboration, evidence, and educational practice?

Supplementary Use of AI

AI added to existing workflowsAI assistance01Prep02Examples03Practice04Feedbackworkflow unchanged — AI as efficiency layer

AI improves efficiency in familiar teaching routines: preparing materials, generating examples, supporting practice, or assisting feedback.

Lesson prepExample generationPractice supportFeedback assist

AI-Native Learning Systems

AI is designed into learning processes, classroom orchestration, collaboration, evidence generation, and teacher decision-making.

AI is not only enhancing education. It is reorganizing learning systems themselves.
Research Agenda

Five questions guiding our long-term research agenda.

Interactive five core research questions Five research questions -- learning, teaching, collaboration, evidence and practice -- evenly orbit AI-native learning systems. Each node is clickable and links to the corresponding research question section. AI-native Learning Systems RQ1: Learning RQ2: Teaching RQ3: Collaboration RQ4: Evidence RQ5: Practice
RQ1

How do AI-native systems reorganize learning?

From individual cognition and static knowledge acquisition to dynamic human–AI learning pathways.

Traditional model
  • Individual cognition
  • Static knowledge acquisition
  • Linear task completion
AI-native model
  • Human–AI synergy
  • Dynamic learning pathways
  • Multi-agent epistemic features
CocoClassCocoStudyCocoFlow
RQ2

How do AI-native systems reorganize teaching?

From lecturer, content provider, and grader to learning environment orchestrator and pedagogical agent designer.

Teacher roles being challenged
  • Delivering content
  • Correcting assignments
  • Managing stable routines
Emerging roles
  • Learning system designer
  • Pedagogical agent creator
  • Interpreter of learning evidence
CocoFlowCocoClassTeacher PA project
RQ3

How do AI-native systems reorganize collaboration?

From human-to-human group work to AI-supported collaboration, regulation, and knowledge building.

Traditional CSCL focus
  • Human interaction
  • Verbal discourse
  • Group cooperation
AI-native collaboration
  • AI as collaborative regulator
  • Socially shared regulation
  • Redistribution of epistemic agency
CocoNoteCollaborative KBMIRACLE
RQ4

How do AI-native systems reorganize learning evidence?

From scores and assignments to process traces, agent logs, collaboration processes, and multimodal evidence.

Past evidence
  • Test scores
  • Quiz results
  • Completed assignments
AI-native evidence
  • Behavioral traces
  • Agent interaction logs
  • Multimodal learning data
CocoViewCocoClassCocoNote
RQ5

How do AI-native systems reorganize educational practice?

From isolated technology adoption to AI systems embedded in real school ecologies and researcher–practitioner collaboration.

Common adoption problems
  • Increased teacher workload
  • Workflow disruption
  • Unsustained pilots
Practice-oriented design
  • Low-friction adoption
  • Integration with school ecology
  • Sustainable research-practice partnership
All platformsSchool pilotsDBR cycles
Intellectual Lineage

Theoretical Foundations

My research is grounded in three intellectual traditions: learning sciences provides the rigor of empirical inquiry and design; posthumanism opens an ontological lens for understanding AI as part of a system’s architecture; and ethical-critical thought reminds me what should not be smoothed away in the name of optimization. They are not parallel sources but a layered, mutually constraining structure of tension.

Empirical & design foundation

Learning Sciences

Constructivism & constructionism (Jonassen · Papert) → Learning Sciences

Learning sciences rejects instructionism: learning is the construction of meaning, not the transmission of knowledge. It inherits constructivism and Papert’s constructionism — understanding deepens when learners build shareable “objects-to-think-with.” Its real signature is a methodological discipline: design-based research treats the real classroom as a site of inquiry, improving practice while generating and refining theory through iterative intervention. This is the thread that carried me from designing tools to designing learning systems — for learning is never an isolated cognitive event, but something that emerges from learners, peers, teachers, tools, and context together.

Meaning construction vs transmissionDesign-based researchMetacognition & self-regulationSocially shared regulationKnowledge buildingOrchestration & scriptingCSCLEvidence-centered design
Relational ontology

Posthumanism

Karen Barad (intra-action) · Deleuze & Guattari (assemblage) · Jane Bennett (vibrant matter)

Posthumanism asks, at the ontological level, what “agency” even is within a system woven from human and non-human actors. Barad’s intra-action holds that entities do not pre-exist and then meet, but emerge through relational entanglement; Deleuze & Guattari’s assemblage treats a learning system as a dynamic configuration of actors, tools, rules, and infrastructure. This lens reveals the limit of the mainstream: “human-in-the-loop” and “human–AI collaboration” seem to protect the human’s place, yet still lock agency back into a logic of human mastery. It supports my view of AI as part of the system’s architecture rather than a tool outside it — and reopens the question of how agency is redistributed across human–machine–world entanglements.

Intra-actionRelational assemblagesDistributed agencyExtended cognitionSocio-technical systems
Ethical & critical counterweight

Critique & Ethics

Emmanuel Levinas (alterity) · Byung-Chul Han (productive friction) · Gert Biesta (the beautiful risk of education)

Decentering the human ontologically must not slide into dissolving ethics — however distributed or optimized a system becomes, the student as Other remains irreducible. Levinas holds that ethics precedes ontology: the “face” of the Other reveals an irreducible alterity that always exceeds my knowledge of them; when a system continuously datafies and predictively “knows” the student, it consumes that alterity through the logic of the Same. Han critiques the culture of “frictionless” optimization — the negativity and friction it removes are precisely the conditions of thinking and learning. Biesta reminds us that the beauty of education lies in its risk. This lens guards what should not be consumed by efficiency, and is the source of my “cognitive friction configuration” construct.

Alterity / irreducibilityThe Same vs the OtherProductive friction / negativityAnti-frictionless optimizationThe beautiful risk of educationCognitive friction configuration

Each plays its part — learning sciences gives the “how,” posthumanism the “what,” ethics and critique the “what must be guarded.” Together they hold one question: when AI enters the very constitution of learning, how are learning, agency, and education itself redefined?

Research Initiatives

Six pathways from research vision to school practice.

Each initiative connects a school-facing problem, a CocoRobo platform, and a research agenda. Two initiatives already have DLW 2026 proposal supporting pages.

DLW 2026 proposal supporting page

Teacher-Created Pedagogical AI Agents

Teachers create, govern, and refine AI teaching agents through structured workflows without coding.

PlatformCocoFlow
Research focusTeacher AI agency · constructionism · AI-TPACK · workflow design
OutcomeTeachers move from AI tool users to AI learning-system designers.
Visit supporting site
DLW 2026 proposal supporting page

AI-Supported Collaborative Learning

Students use AI-supported shared workspaces to build knowledge, regulate collaboration, and learn together.

PlatformCocoNote
Research focusCSCL · knowledge building · socially shared regulation
OutcomeAI supports collaboration, not only one-to-one tutoring.
Visit supporting site
Ongoing / Pilot

Student AI Literacy Curriculum & Assessment

Schools build structured AI literacy learning experiences and assessment frameworks for students.

PlatformCocoClass · CocoStudy
Research focusAI literacy · assessment design · evidence-centered learning
OutcomeStudents understand, use, evaluate, and govern AI more responsibly.
Ongoing / Pilot

AI Agents for Student Learning Enhancement

AI agents embedded in classroom orchestration support practice, feedback, and adaptive learning decisions.

PlatformCocoClass × CocoFlow
Research focusAgent-supported learning · feedback · classroom orchestration
OutcomeTeachers can personalize support while keeping control of learning flow.
Ongoing

Classroom Discourse Analytics for Teacher Reflection

Classroom talk and interaction traces become evidence for teacher reflection and instructional improvement.

PlatformCocoView
Research focusDiscourse analytics · multimodal evidence · teacher reflection
OutcomeSchools can see classroom processes that are usually invisible.
Ongoing

AI-Supported Self-Regulated Learning

Students receive personalized practice, diagnostic feedback, and learning pathway support beyond the classroom.

PlatformCocoStudy
Research focusSRL · adaptive learning · longitudinal learning analytics
OutcomeLearning support continues across lessons, practice, and reflection.
Research Platforms

Products as research platforms, not isolated tools.

Each platform is designed for real classroom use while making learning processes observable, interpretable, and improvable.

F

CocoFlow

For schools: teachers can create AI teaching agents without coding.

Research focus: teacher agency, human-AI co-design, multi-agent orchestration.
Observable: workflow design, agent configuration, boundary decisions.
C

CocoClass

For schools: PPT-based teaching becomes interactive, adaptive, and evidence-rich.

Research focus: classroom orchestration, low-friction AI adoption, real-time analytics.
Observable: teacher orchestration, student responses, classroom interaction traces.
S

CocoStudy

For schools: students receive practice, feedback, and learning support beyond class time.

Research focus: SRL, adaptive learning, personalized AI support.
Observable: learning pathways, error patterns, feedback cycles, self-regulation traces.
N

CocoNote

For schools: students collaborate through shared artifacts, inquiry, and AI-supported regulation.

Research focus: CSCL, knowledge building, SSRL, relational epistemic agency.
Observable: collaborative discourse, shared artifacts, regulation patterns.
V

CocoView

For schools: classroom discourse and interaction become visible for teacher reflection.

Research focus: discourse analytics, multimodal evidence, teacher reflection.
Observable: classroom talk, multimodal interaction, reflection evidence.
School Deployment

CocoRobo serves 1,400+ schools across its full product range

Each platform above is live in real classrooms across Hong Kong, Macau, and the Greater Bay Area, forming a complete AI-native learning infrastructure. For school-facing product details, the SMART framework, and teacher onboarding guides, visit the page designed for principals and teachers.

CocoRobo SMART — School Product Overview ↗
Research Network

A hybrid research-practice platform.

CocoRobo Research brings together product design, learning sciences, AI system development, classroom implementation, and academic publication.

CocoRobo Research Team

A hybrid team combining learning designers, product researchers, engineers, school implementation teams, and research collaborators.

Long-term university collaborators

Sustained research collaborations with leading universities across Hong Kong and Mainland China.

The Chinese University of Hong Kong
The University of Hong Kong
The Education University of Hong Kong
South China Normal University
Nanjing Normal University
Methodology

Design-Based Research in real educational ecologies.

CocoRobo Research adopts iterative design-based research: design, implementation, data analysis, reflection, and redesign with schools, teachers, students, researchers, and AI systems.

Multimodal learning analytics Discourse analysis Process mining AI interaction traces Classroom observation Interviews Longitudinal implementation
Design-based research loop An infinity-shaped design-based research loop connects design, implement, analyze, and reflect & iterate, with classroom data, school ecology, and AI-native learning as the object of study at the center. Design Implement Analyze Reflect & Iterate Object of study Evidence streams interaction logs · discourse · classroom traces AI workflow data · observation · reflection notes Real educational ecologies schools · teachers · students · platforms
Selected Research Outputs

Research outputs.

11 selected research outputs across AERA, ISLS, AIED, and CSCL, including the CSCL 2025 Outstanding Short Paper Award.
2026 AIED1 paper

Computational Thinking Development in AI Agent Creation: A Mixed-Methods Study

Sun, Y., Xin, H., Niu, Q., Li, S., Huang, L., & Chen, G. (2026, June). Computational thinking development in AI agent creation: A mixed-methods study [Short paper]. Proceedings of the 2026 International Conference on Artificial Intelligence in Education (AIED), Seoul, Republic of Korea.

AIED 2026Short paperStudent development
arXiv ↗
2026 ISLS4 outputs

An Activity-Theoretical Approach to Teacher Professional Development in Pedagogical AI Agent Design

Xin, H., Niu, Q., Li, S., Sun, Y., Chai, C., Huang, L., & Chen, G. (2026, June). An activity-theoretical approach to teacher professional development in pedagogical AI agent design [Long paper]. Proceedings of the 2026 ISLS Annual Meeting, Irvine, CA.

ISLS 2026Long paperCocoFlowTeacher agency
arXiv ↗

Modeling AI-TPACK in Practice: Insights from Teachers’ Multi-Agent Workflow Design

Sun, Y., Xin, H., Li, S., Niu, Q., Chai, C., Huang, L., & Chen, G. (2026, June). Modeling AI-TPACK in practice: Insights from teachers' multi-agent workflow design [Short paper]. Proceedings of the 2026 ISLS Annual Meeting, Irvine, CA.

ISLS 2026Short paperAI-TPACKMulti-agent workflow
arXiv ↗

MIRACLE: Multi-Agent Intelligent Regulation to Advance Collaborative Learning Environment

Li, S., Xin, H., Sun, Y., Niu, Q., Chai, C., Huang, L., Chen, G., & Zhang, Y. (2026, June). MIRACLE: Multi-agent intelligent regulation to advance collaborative learning environment [Short paper]. Proceedings of the 2026 ISLS Annual Meeting, Irvine, CA.

ISLS 2026Short paperMulti-agent systemsCollaborative learning
arXiv ↗

Hierarchical Multi-Agent System for Instructional Design in Music Knowledge Building

Zhang, M., Zhu, M., Xin, H., & Zhang, Y. (2026, June). Hierarchical multi-agent system for instructional design in music knowledge building [Poster]. Proceedings of the 2026 ISLS Annual Meeting, Irvine, CA.

ISLS 2026PosterKnowledge buildingInstructional design
2026 AERA2 outputs

Teachers’ Behavior in Building Agents Based on Hierarchical Clustering and Thematic Analysis

Xin, H., Yu, Y., Li, S., Niu, Q., Gao, L., Huang, L., & Chai, C. (2026, April). Teachers' behavior in building agents based on hierarchical clustering and thematic analysis [Roundtable]. 2026 AERA Annual Meeting, Los Angeles, CA.

AERA 2026RoundtableTeacher agent design

Empowering Teachers as Creators of Pedagogical Agents: An Integrated Perspective of Constructionism, ICAP, and TPACK

Li, S., Xin, H., Yu, Y., Niu, Q., Gao, L., Huang, L., & Chai, C. (2026, April). Empowering teachers as creators of pedagogical agents: An integrated perspective of constructionism, ICAP, and TPACK [Poster]. 2026 AERA Annual Meeting, Los Angeles, CA.

AERA 2026PosterConstructionismICAPTPACK
2025 ISLS / CSCL2 outputs
🏆 CSCL 2025 Outstanding Short Paper Award

CocoNote: Agents-Aided Collaborative Learning Environment Enhances Socially Shared Regulation

Xin, H., Li, S., Huang, L., Yip, V. W. Y., Niu, Q., Chen, X., & Liu, J. (2025). CocoNote: Agents-aided collaborative learning environment enhances socially shared regulation [Short paper]. Proceedings of CSCL 2025. Outstanding Short Paper Award.

CSCL 2025Short paperCocoNoteSSRLOutstanding Short Paper Award
ISLS ↗

A Multi-Agent System (MAS)-Based Tool to Support Novice Teachers in Knowledge Building Pedagogy

Xin, H., Lan, L., Niu, Q., Hu, Z., & Zhang, Y. (2025). A multi-agent system (MAS)-based tool to support novice teachers in knowledge building pedagogy . Proceedings of CSCL 2025.

CSCL 2025Multi-agent systemsKnowledge buildingTeacher support
ISLS ↗
2024 ISLS / Other2 outputs

CocoNote Supported Project-Based Learning Environment: Perspectives of Construction and Collaboration

Xin, H., Niu, Q., Lan, L., Xiao, Z., & Wu, F. (2024). CocoNote supported project-based learning environment: Perspectives of construction and collaboration [Poster]. Proceedings of CSCL 2024.

CSCL 2024PosterCocoNotePBLCollaboration
ISLS ↗

The Design and Application of RAG-Based Conversational Agents for Collaborative Problem Solving

Zhong, X., Xin, H., Li, W., Zhan, Z., & Cheng, M. (2024). The design and application of RAG-based conversational agents for collaborative problem solving . Proceedings of the 2024 9th International Conference on Distance Education and Learning.

2024RAGConversational agentsCollaborative problem solving
Research Notes & Perspectives

A public archive of CocoRobo Research thinking.

Notes connect product practice, school observations, and broader reflections on AI-native education.

Founder’s Perspective

Reintroducing CocoRobo: What are we building when AI redefines learning, capability, and education itself?

This essay frames CocoRobo Research around AI-native learning systems, human-AI collaboration, cognitive sovereignty, teacher transformation, and AI-native learning infrastructure.

AI-native learning systemsCognitive sovereigntyLearning infrastructure
Read original WeChat article ↗
CocoRobo Research Note · Observation

Beyond the Ministry report: when 86% of teachers worry students may lose independent thinking

This essay connects national-level teacher AI findings with classroom stories: withdrawing AI’s evaluation authority, setting timing constraints, making AI deliberately “less smart,” and teachers learning to “teach AI.”

Teacher AI agencyAI governanceCocoClassCocoFlow
Read original WeChat article ↗
CocoRobo Research Note · Field

When students build AI Agents for their parents and grandparents — and learn to let AI step back

An agent-building summer camp where 100+ students designed working AI Agents for real people in their lives. The essay shows how student AI literacy “grows,” and how they learn to let AI step back at the right moments across relational, cognitive, authenticity, and power boundaries.

Student AI literacyLetting AI step backCocoFlow
Read original WeChat article ↗
CocoRobo Research Note · Subject exemplar

How to bring AI into a subject: a hands-on example from primary science

Using primary science as an example, a systematic account of the knowledge and methods for bringing AI into a subject — the design thinking behind the free interactive courseware.

Subject exemplarAI-enabled teachingDesign thinking
Read original WeChat article ↗
Human AI Environment Learning Activity (emergent)

Our object of study is not AI alone.
It is the emerging relational structure among humans, AI, environments, and learning activity.

School Stories & Impact

What AI-native learning systems look like in schools.

These scenarios show the bridge between research concepts and daily educational practice.

Teacher

A teacher designs an AI agent for a lesson

Instead of asking AI to generate generic materials, the teacher defines learning goals, student level, task boundaries, and feedback rules.

Result: teacher control is preserved while AI reduces preparation and feedback burden.
Students

Students use CocoNote for collaborative inquiry

AI does not replace group discussion; it helps students organize ideas, regulate collaboration, and build shared artifacts.

Result: collaboration becomes visible, discussable, and improvable.
School

A school uses classroom evidence for reflection

Classroom talk, student responses, and interaction traces become evidence for lesson improvement and teacher professional learning.

Result: school-based AI innovation becomes evidence-informed rather than tool-driven.
Practice Cases

Classroom and teacher-design cases behind the research.

Selected classroom and teacher-design cases for teachers and school leaders. Open each source article to learn more.

Teaching Agents

From automation to thinking support

Teaching agents are positioned differently from task-completion agents: they pause, ask, scaffold, and help students think rather than simply finishing the work for them.

In school: Use this when schools worry that AI will replace student thinking.
Research lens: Teacher-governed AI · cognitive scaffolding · agency
Read source article ↗
English Classroom

A triadic teacher–student–AI English lesson

In an English lesson on “Our lives in the future,” the platform and AI agents supported pre-reading, drafting, feedback, and revision across the lesson flow.

In school: Use this for schools seeking a concrete teach–learn–assess integration case.
Research lens: Classroom orchestration · formative assessment · AI feedback
Read source article ↗
Classroom Visibility

From blind-box marking to classroom-wide visibility

The teacher dashboard makes student submissions and AI interaction records visible, helping teachers identify who needs support during class.

In school: A practical entry point for schools wanting visible learning evidence.
Research lens: Learning analytics · classroom monitoring · evidence generation
Read source article ↗
View more cases 9 additional classroom and teacher-design examples
Teacher Design

Design tutor + testing assistant for teacher-created agents

A two-agent workflow helps teachers move from an initial teaching idea to a tested pedagogical agent, including boundary checks before students use it.

In school: Start with one teacher, one lesson, and one agent-testing cycle.
Research lens: Teacher AI agency · boundary design · human-in-the-loop
Read source article ↗
Teach–Learn–Assess

From delayed marking to a live learning loop

Student writing is submitted digitally, reviewed by AI, revised by students, and monitored by the teacher, turning evaluation into a dynamic learning process.

In school: Start with one writing task and use AI to support first-round revision.
Research lens: Feedback loop · metacognition · self-regulated learning
Read source article ↗
Vibe Coding

Interactive HTML learning tools for abstract concepts

Teachers use AI-generated HTML tools to turn abstract ideas such as algorithms, binary code, and decision-making into playable and manipulable learning experiences.

In school: Useful for ICT, STEM, maker education, and interdisciplinary lessons.
Research lens: Computational thinking · embodied interaction · low-threshold creation
Read source article ↗
Teacher Innovation

From technical implementation to learning experience design

Vibe coding shifts teachers’ attention from writing code to designing meaningful interactions, scenarios, simulations, and student decision points.

In school: Teachers can start from a prompt and gradually build a reusable school-based resource bank.
Research lens: Teacher creativity · design thinking · AI-assisted production
Read source article ↗
Real-Time Feedback

AI teaching assistant for classroom writing feedback

A writing-feedback agent turns delayed marking into real-time support, allowing students to revise while their thinking process is still active.

In school: Use this when teachers face large-class feedback pressure.
Research lens: Formative feedback · ICAP · writing analytics
Read source article ↗
Argumentation

AI debate room for building high-quality mental models

Students observe AI agents debating both sides of an issue, then evaluate the reasoning process instead of passively memorizing argument rules.

In school: Suitable for Chinese, history, science, citizenship, and cross-curricular thinking lessons.
Research lens: Argumentation · critical thinking · epistemic cognition
Read source article ↗
Critical Thinking

Toulmin argument tutor for claim–evidence reasoning

A structured argument agent guides students through claims, evidence, warrants, counterarguments, and self-evaluation.

In school: Use this for essays, debate preparation, and historical explanation tasks.
Research lens: Toulmin model · cognitive scaffolding · procedural knowledge
Read source article ↗
History Inquiry

Festival research expert for personalized historical inquiry

Students choose a traditional festival, question an AI research expert, and build a timeline of origin, development, and cultural transmission.

In school: A low-risk entry point for history inquiry and student questioning.
Research lens: Inquiry learning · personalization · student agency
Read source article ↗
Multi-Agent Workflow

Activity-planning assistant for complex project tasks

A workflow-based agent decomposes a complex activity-planning task into steps such as theme, purpose, content, division of labor, budget, materials, and safety plan.

In school: Useful for project-based learning, interdisciplinary activities, and student leadership tasks.
Research lens: Workflow scaffolding · PBL · procedural knowledge
Read source article ↗
Collaborate

How collaboration usually starts.

We work with schools, education bureaus, universities, foundations, and international organizations to design, study, and improve AI-native learning systems.

For Schools & Education Leaders

What schools can do with CocoRobo Research.

We help schools move from trying isolated AI tools to building classroom-ready, teacher-governed, research-informed AI learning systems.

01

Create AI teaching agents without coding

Teachers can design pedagogical AI agents through structured workflows, not programming.

Research anchor: teacher AI agency · constructionism · AI-TPACK
02

Make classroom learning visible

Capture student responses, discourse, collaboration patterns, and learning traces.

Research anchor: learning analytics · multimodal evidence
03

Support students beyond one-to-one AI tutoring

Use AI to support collaboration, inquiry, regulation, AI literacy, and self-regulated learning.

Research anchor: CSCL · SSRL · knowledge building
04

Build school-based AI innovation with research support

Start pilots, teacher development programmes, and evidence-rich classroom cases.

Research anchor: design-based research · school ecology
Collaboration steps
01

Exploratory conversation

Understand school goals, grade levels, subject needs, and current AI readiness.

02

Pilot design

Choose an initiative: teacher agents, collaboration, AI literacy, classroom analytics, or self-regulated learning.

03

Teacher onboarding and implementation

Keep workload low with platform support, templates, and research-informed implementation design.

04

Evidence, reflection, and scaling

Review learning evidence, teacher feedback, and decide whether to scale, publish, or extend.

What schools may receive

Platform access during pilotTeacher professional learningClassroom implementation supportLearning analytics or research reportSchool-based case developmentPossible research collaboration under proper ethics and consent procedures

What schools usually need to commit

One school coordinator or lead teacherOne pilot class, subject, or grade-level groupA short teacher onboarding sessionAgreement on data, privacy, and ethics boundariesTime for reflection and feedback after implementation

Founder profile

For Tony Xin’s personal background and longer-form reflections on AI-native education, visit his personal website.

Visit haiyang.xin ↗
Research ethics. Project pages should include project-specific ethics statements. Where research involves students, teachers, classroom records, or identifiable data, ethics approval and privacy protection measures should be stated clearly.
Contact CocoRobo Research