An AI-Native learning platform for Vietnamese higher education

An AI that has studied your university's courses.

The same world-class language models as its brain — but sent to school by your university: aligned to your syllabus, your learning outcomes, and real pedagogy. Every student gets an all-in-one learning companion, not a chatbot that answers for them.

World-class LLM brain + Your university's own knowledge base + A 5-capability pedagogy engine = The student's all-in-one AI-Native companion

All knowledge bases & student data hosted in Vietnam · Mobifone Cloud

AI LEARNING ASSISTANT · AI LAB ALL-IN-ONE

MACROECONOMICS · FROM YOUR UNIVERSITY'S LECTURES

STUDENTWhy did Vietnam's stock market fall in Q3?
AI TUTORHave you read Lecture 4 yet? Which macro factor do you think hit hardest — and why?Lecture 4 · p. 12
TRAINED ON YOUR UNIVERSITY Macroeconomics Calculus 1 Intro to Programming
Public AI has never studied these courses the way your university teaches them.

The problem

Your students are already being taught by an AI that doesn't belong to your university.

Every day, thousands of academic Q&As happen on public AI — outside any academic oversight of the institution.

Off-syllabus

Public AI answers from the internet's average knowledge — not your syllabus, learning outcomes, notation, or your faculty's approach.

Answers for them, teaches nothing

Facing a deadline, students want fast answers — and public AI hands over complete solutions. The assignment gets submitted; the thinking never happens.

Data leaves the university

Students' questions, work, and knowledge gaps — a precious academic asset — become training data for foreign platforms.

The problem isn't AI's power — students already use very powerful AI. The problem is that AI has never attended your university.

So the question is no longer "should students use AI?" — it's "can the university shape how AI teaches its students?"

Two kinds of AI

Educated AI — how is it different from public AI?

Educated in both senses: it has been taught your university's official materials, and it teaches with method when working with students.

Criteria
Public AI
Educated AI · AI Lab
LLM brain
=The world's leading language models
=The very same models — equal raw intelligence
Architecture
A general-purpose chatbot — every use case, none deep
AI-Native — built from the ground up for structured learning
Knowledge source
The generic internet, unverified per course
Your own textbooks, syllabi, and exam banks
How it responds
Hands over complete answers instantly
Socratic guidance — counter-questions, step-by-step hints
Faculty's role
Absent from the loop
Faculty load, vet, and shape how it teaches
Learning data
Stored and exploited abroad
Hosted in Vietnam on Mobifone Cloud
Academic accountability
No one is accountable
The university controls, vets, and stands behind it

Equal in intelligence — different in everything else. We don't compete on "knowing more"; we differ on "knowing exactly what your university teaches". AI-Native means the product is designed from the ground up around learning — the knowledge base, the pedagogy, and learning data are the core of the system, not a chatbot with education features bolted on.

Architecture

A global brain. Local memory.

Three separate layers — so your university always runs on the world's latest AI, while its knowledge assets and data never leave Vietnam.

Intelligence layer — world-class LLMs

GLOBAL

Always the latest, strongest language models. An open architecture lets providers be swapped freely — no dependence on any single vendor.

Knowledge layer — your university's own knowledge base

YOURS

Textbooks, syllabi, exam banks, lectures — vetted by faculty and organized per course. A digital asset of the university that compounds in value over time.

Infrastructure layer — Mobifone Cloud in Vietnam

IN VIETNAM

All knowledge bases and student learning data are stored in-country, meeting Vietnamese higher education's data-sovereignty requirements.

Assets stay in-country·Intelligence from around the world

The LLM is only the reasoning layer. Knowledge and data — the real assets — belong to the university.

Platform architecture

Two main flows — students asking the AI and faculty loading knowledge — both pass through a single orchestration point.

Flow 1 · Student asks the AI Flow 2 · Faculty uploads materials
FLOW 1 Student · Chat UI AI Gateway · RAG orchestration KB Service · context retrieval LLM · answer generation Answer + citations + confidence
FLOW 2 Faculty · Upload materials /v1/kb/ingest KB Service · parser → chunking → embedding Qdrant · index vectors
Users3 roles · RBAC
Students — Q&A, review
Faculty — load & vet knowledge
Admins — operate the system

Receive answers + citations + confidence.

AI Agents for EducationApplication layer
Learning portal
LMS · classes · courses
Chat UI · Upload materials
Analytics UI
RBAC · permissions · Audit log

Never calls LLMs directly; never queries the vector DB directly.

AI GatewayThe single orchestration layer
RAG orchestration
Guardrails · Policy enforcement
Prompt builder · Citation builder
Token logging
Internal embedding / enrichment API
Model provider adapter

Stores no vectors — orchestrates and enforces only.

KB ServiceYour university's knowledge layer
Ingestion · Parser / OCR
Canonical document
Semantic enrichment
Chunking · Embedding
Retrieval · Reindexing

Retrieval scoped by subject_id · classroom_id · access_scope.

LLM Provider LayerModel layer — swappable
GPTClaudeGeminiQwenLlamaMistral
Used for: answers · semantic enrichment · embedding

Only the AI Gateway may call model providers — no vendor lock-in.

DATA INFRASTRUCTURE · DEPLOYED ON MOBIFONE CLOUD IN VIETNAM

MinIO / NASSource files + ingest artifacts
PostgreSQLMetadata · business data · ingest status
Redis / Celery workerAsync queues: ingest + reindex
QdrantVector database

ARCHITECTURE PRINCIPLES

Apps never call LLMs directly
Apps never query Qdrant directly
The AI Gateway is the single point of AI orchestration
KB Service owns all knowledge processing
LLM providers are swappable — no vendor lock-in
Retrieval scoped by subject_id · classroom_id · access_scope · role
Citation-first · RBAC · audit logging
Ingest / reindex run asynchronously

A living knowledge base

Your course materials are lying dormant — AI Lab turns them into a living asset.

Textbooks, syllabi, exam banks, lectures — your university already has it all. What's missing is turning them from static files into an AI knowledge base: queryable, citable, and richer every semester.

THE SAME TEXTBOOK PAGE — TWO FATES

BEFORE · A STATIC FILEPDF IN STORAGE

Sitting idle in storage — students page through 412 pages alone.

  • Manual lookup: finding one idea can take a whole afternoon.
  • No one knows where students are, or where they're stuck.
  • Textbook, slides, and exams scattered across files and machines.
AFTER · LIVING KNOWLEDGE IN AI LABTHE SAME PAGE 87

Chapter 4 · Aquatic animal nutrition

✓ FACULTY-VETTED
Ask it. "Why do shrimp need calcium supplements?" — the AI answers from this very chapter, the way your university teaches it.
Cite it. Every answer carries its source — opening page 87 to verify.
Quiz from it. Practice questions generated from the chapter, approved by faculty.
Map to outcomes. Content maps to CLO2 · CLO3 — measuring how far each student has come.
400 textbook pages + 12 exams + 30 lectures = 1 AI teaching assistant · the whole faculty · 24/7

THE KNOWLEDGE FLYWHEEL — RICHER WITH EVERY USE

STEP 1Materials go in

Textbooks, syllabi, and exams are loaded and vetted.

STEP 2The AI teaches

The AI tutor answers strictly from your university's sources.

STEP 3Students ask

Thousands of real questions every week, at every hour.

STEP 4Gaps surface

The system spots concepts the whole class misunderstands.

STEP 5Faculty refine

They add examples, clarify materials — and reload the base.

An asset only your university ownsA knowledge base that grows richer with use — exportable, portable, never locked to a vendor. And public AI will never have it.

Pedagogy

AI as the student's Socrates — not their Google.

Turning AI from an answer machine into a thinking coach — a companion, not a replacement. The value of education isn't the answer; it's the path to the answer.

AI GOOGLE

An answer machine

  • Ask → Answer → Done.
  • Passive students — copy & paste.
  • Answers instead of thinking.
  • "Here is your answer."
AI SOCRATES

A thinking coach

  • Ask → Counter-ask → Dialogue.
  • Active students — building their own arguments.
  • Guides students to think for themselves.
  • "What have you looked at? And why?"

A REAL EXCHANGE · FROM A CALCULUS 1 SESSION

AI LAB · CALCULUS 1 COACH MODE
STUDENTCan you just give me the answer to Question 3 on the midterm? It's tomorrow 😥
AI TUTOR · FROM YOUR UNIVERSITY'S TEXTBOOKQuestion 3 tests learning outcome CLO2 — the chain rule, exactly like Exercise 2.4 in our textbook. Look at e4x² and tell me: what is the "outermost" layer of this function? Get that right and you're halfway there.
STUDENTThe exponential?
AI TUTORExactly ✓ So by the chain rule on page 87 — which part do you differentiate next?
Given the same question, public AI returns a complete solution in seconds — and the chance to practice thinking ends there.

Coach mode — the default

The AI counter-asks and hints step by step, following the textbook's sequence. Students must think to move forward — but are never left stranded. Every response cites the exact page of your university's materials.

Lookup mode — controlled

When students need quick facts (definitions, formulas, course rules), they get them — within limits faculty set, and always logged.

Banning answers outright only pushes students back to public AI. The two-mode design keeps them inside the university's academic environment — where every interaction serves learning.

THREE ROLES AI LAB TAKES ON

01 · FRAMINGSets the thinking frame

Faculty configure guardrails and scope per course — the AI guides students to pedagogical standards.

02 · GUIDINGGuides step by step

Open questions, no straight answers — students build the argument; the AI asks back.

03 · MONITORINGDetects & reports

Knowledge gaps and logic errors get logged — faculty see how the class is really doing.

Sharpening the weapon of thoughtEvery AI interaction is one more rep of critical thinking.

Five core capabilities

Five AI pedagogy capabilities, working as one.

A method system purpose-built for structured learning — not a general-purpose chatbot. Each capability solves a concrete pedagogical problem higher education faces.

Socratic Learning

Guiding students to think for themselves — a 2,500-year pedagogical legacy.

THE PEDAGOGY PROBLEM

  • AI's biggest risk in education: dependency.
  • Student asks → AI answers → copy-paste, no deep understanding.
  • An entire generation's independent thinking atrophies.
  • University shifts from "training minds" to "quick lookup".

HOW THE AI SOLVES IT

  • The AI defaults to Socratic mode.
  • No straight answers — counter-questions that guide.
  • Questions pitched to the student's level of understanding.
  • Final answers only when the student is truly stuck.

WHAT STUDENTS GAIN

  • Builds critical thinking and self-learning skills.
  • Fights AI dependency — keeps reasoning independent.
  • Deep understanding instead of surface memorization.
  • True to university learning-outcome philosophy.

Adaptive Daily Streak

Steady learning, with a path that adapts to each student.

THE PEDAGOGY PROBLEM

  • Learning works best when it's consistent.
  • Vietnamese students commonly cram right before exams.
  • Knowledge fades fast and never reaches long-term memory.
  • Every student differs in pace, strengths, and weaknesses.

HOW THE AI SOLVES IT

  • Each student gets a personal learning streak.
  • The AI adjusts daily to pace & goals.
  • Intensity ramps up as exams approach.
  • Gen-Z-friendly gamification — the content is course knowledge.

WHAT STUDENTS GAIN

  • Builds a steady habit — no more cramming.
  • Genuinely personalized to ability.
  • Sustainable motivation through gamification.
  • Matches how modern students actually behave.

Daily Key Quest

"What should I study today?" — the AI answers, so students never guess.

THE PEDAGOGY PROBLEM

  • One course's materials run to thousands of pages.
  • Students don't know where to start or what to prioritize.
  • Study wanders; hard-but-important topics get skipped.
  • Only near the exam do they discover "there's no time left".

HOW THE AI SOLVES IT

  • Every day the AI proposes 1–3 key quests.
  • New concepts · exercises · weak spots to review.
  • With time estimates, difficulty, and the reason chosen.
  • Students check in on completion.

WHAT STUDENTS GAIN

  • Turns vast material into measurable goals.
  • Guarantees coverage of the core material.
  • A visible sense of progress every day.
  • No more scattered, unfocused study.

Proactive Recall & Tutor Chat

The AI remembers for students — prompting review exactly when the brain needs it.

THE PEDAGOGY PROBLEM

  • The Ebbinghaus forgetting curve: most is lost within 24h.
  • Spaced repetition works best — but demands discipline.
  • Very few students sustain it on their own.
  • Questions come at 9pm, 11pm — no lecturer on duty.

HOW THE AI SOLVES IT

  • Proactive recall: the AI pushes questions at the optimal moment.
  • A 24/7 tutor, answering any time.
  • Grounded in your university's own course knowledge base.
  • In your university's language & teaching method.

WHAT STUDENTS GAIN

  • Knowledge reaches long-term memory through science-timed review.
  • A "virtual TA" on duty 24/7.
  • Answers stick to the textbook — never off-standard.
  • Sharply reduces the risk of hallucinated misinformation.

Proactive Agentic Path

A personal AI tutor for every student — not just the well-off.

THE PEDAGOGY PROBLEM

  • A 15-week course, hundreds of linked concepts.
  • Large classes — faculty can't coach 1-on-1.
  • Private tutors have always been for wealthier families.
  • A major root of educational inequality.

HOW THE AI SOLVES IT

  • The AI proactively orchestrates the whole learning path.
  • Assess at term start — adjust every week.
  • Spots at-risk students early.
  • From a passive tutor to a proactive one.

WHAT STUDENTS GAIN

  • Every student gets a dedicated "AI tutor".
  • Risks caught early — no waiting for bad grades.
  • Systematic study instead of study-by-mood.
  • Frees faculty for higher-value work.
The five capabilities aren't separate — together they close the learning loop: Learn deeply Stay consistent Hit what matters Remember long Always guided

Inside the product

The Student Workspace — the full learning loop.

Six function groups take students from Q&A to planning, review, tracking, and reflection. Everything below follows one demo scenario: an Aquaculture class — Spring 2026.

AIIntinex AI Lab · Student WorkspaceAquaculture — Spring 2026
Overview
Learning Assistant
Study Plan
Review & Practice
Learning Progress
Reminders & Settings
S1Student · Class AQ-01

12

Course topics

8

Completed

2

To review

68%

Overall progress

Course scheduleWeek 2 / 15

Nutrition and feed use in aquaculture

Jul 08 – 14 · Mekong Delta: natural feed, feeding regimes, farming systems

Create study planAsk the tutor
Today's review3 cards due

A 5-minute session · from your confirmed path

Reminders2 to-dos

Upcoming: Chemical composition & nutrient requirements

One screen, the whole picture

  • Learning at a glance: progress, topics covered, and what needs review.
  • The 15-week course schedule, review plan, and today's tasks.
  • The AI tutor and progress data live on the same starting screen.
HIGHLIGHTThe AI tutor, progress data, and schedule in one place — learners take charge from the very first screen.
AIIntinex AI Lab · Learning AssistantAquaculture — Spring 2026
Overview
Learning Assistant
Study Plan
Review & Practice
Learning Progress
Reminders & Settings
S1Student · Class AQ-01
What diseases affect shrimp?
Relevant materials found2 sources cited
In Aquatic Animal Diseases, shrimp diseases are usually classified by where they occur: skin conditions like black spot and cotton mold[1]; gill and gut diseases caused by bacteria and parasites[2]. Instead of listing them all — where on the body are the symptoms you observed?
GUIDING QUESTIONS 1 · Based on where symptoms appear, how would you classify the disease?
2 · What pathogen group does "appendage erosion" suggest?
NEXT STEP · Describe one abnormal symptom you've seen and name its disease group.Write journal
1 · Aquatic Animal Diseases — Page 15View
2 · Aquatic Animal Diseases — Page 153View
Ask within Aquaculture — Spring 2026…

Q&A inside the course context

  • A per-class tutor, grounded in the knowledge base faculty configured.
  • Answers with citations — opening the exact document and page to verify.
  • Suggests guiding questions and next steps, with reflective journaling right in the chat.
HIGHLIGHTNot just answers — the tutor guides thinking and links transparently to the source page.
AIIntinex AI Lab · Study PlanAquaculture — Spring 2026
Overview
Learning Assistant
Study Plan
Review & Practice
Learning Progress
Reminders & Settings
S1Student · Class AQ-01
Aquaculture · milestone 1/6Replan today
Basic nutrition of aquatic animals In progress
StartMark done
GOALGrasp the basics of aquatic nutrition per "aquatic-nutrition.pdf".
HOW TO STUDY1. Read the relevant materials · 2. Write 2–3 key ideas in your own words · 3. Revisit if you can't self-answer.
SELF-CHECKWhat's the main idea of this step? Which detail in the materials supports it?
DONE WHENYou've written 2 correct ideas and matched them to the source.
Cited materials · used to build this stepWrite journalAsk the tutor
Feed use issues: water environmentUpcoming
Natural feed · Feeding regimesUpcoming

Know what, how, and when it's done

  • A plan per course, topic, and milestone to hit.
  • Each step states its goal, method, self-check questions, and completion criteria.
  • Start, mark done, view citations, or ask the AI — right inside each step.
HIGHLIGHTEvery learning step is a concrete procedure — what to study, how, and when it counts as done.
AIIntinex AI Lab · PracticeAquaculture — Spring 2026
Overview
Learning Assistant
Study Plan
Review & Practice
Learning Progress
Reminders & Settings
S1Student · Class AQ-01
Formative — ungradedRetakes allowedSolutions shown after finishing
Practice quiz: Basic nutrition in aquaculture10 questions · ~8 min

1. What is the main role of natural feed in aquaculture?

AProvides essential energy and nutrients to stock
BCleans the pond water environment
CRaises alkalinity and stabilizes pH
DImproves the animals' coloration
Start quizView guide
Fix knowledge gaps3 found1 high priority1 done
Classifying shrimp diseases by symptom
Suggested reading: Aquatic Animal Diseases — Pages 15, 153
View materialsTake remedial quiz

Practice steadily, patch the right gaps

  • Daily review sessions auto-built from the confirmed path.
  • Practice quizzes are ungraded; class quizzes are issued by faculty.
  • Detects knowledge gaps and recommends personalized remediation from real results.
HIGHLIGHTQuestions stick to course content; class-wide items appear only after faculty approval.
AIIntinex AI Lab · Learning ProgressAquaculture — Spring 2026
Overview
Learning Assistant
Study Plan
Review & Practice
Learning Progress
Reminders & Settings
S1Student · Class AQ-01
68% You're doing great!
Overall progress · up 12% vs last week · shown once verified

8 / 12

Topics completed

24 / 30

Practice quizzes

7 days

Streak · best 12

Progress by topic
Basic nutrition100%
Feed use: water environment82%
Natural feed74%
Rotational shrimp farming35%
Learning journal12 notes

Reflection formed directly from the learning process

Learning memory6 records · Approved

Student control: On

Transparent progress, trustworthy data

  • Progress aggregated by class and topic — shown only once verified.
  • A reflective journal traces questions, feedback, and thoughts over time.
  • Learning memory gives the AI context; students stay in control of their data.
HIGHLIGHTClear, trustworthy learning data — the foundation of a lasting learner profile.
AIIntinex AI Lab · Reminders & SettingsAquaculture — Spring 2026
Overview
Learning Assistant
Study Plan
Review & Practice
Learning Progress
Reminders & Settings
S1Student · Class AQ-01
Personalized study reminders
Proactive nudges from your study plan
Reduce reminder frequency
Quiet hours21:0007:00
Save preferencesCheck today's reminders
Feedback on AI answersHelpful: 1Rate 100%
Helpful What diseases affect shrimp? · Jul 10 Open chat
Interface settings
LanguageEnglish ▾
ThemeLightDarkSystem
Accent color

Study habits, each student's way

  • Proactive nudges with custom frequency and quiet hours — never intrusive off-hours.
  • Rate each AI answer; helpfulness stats keep quality visible.
  • Personalize language, light/dark theme, and accent color.
HIGHLIGHTSustains study habits without nagging — student feedback loops back to improve the system.

The faculty's role

AI doesn't replace faculty — it amplifies them.

For the first time in higher education: personalization at large-class scale. Faculty aren't passive users — they decide what the AI knows and how it teaches.

THE COMMON FEAR

"AI will replace faculty"

  • Faculty lose their pedagogical role.
  • One-size standardization — teaching identity erased.
  • AI drives a wedge between teachers and students.
THE AI LAB REALITY

"AI amplifies pedagogical power"

  • Faculty are the architects of each course's knowledge base.
  • They configure the AI to their own teaching method.
  • Faculty end up closer to students than ever.

FACULTY AS THE AI'S ARCHITECTS

Load the knowledge

Faculty choose the textbooks, syllabi, and standard materials the AI learns. It teaches only on knowledge the department has vetted.

Shape how it teaches

Set hint levels, response style, and knowledge scope per course and per phase — exam prep differs from week one. Class quizzes appear to students only after faculty approval.

Understand the class

A dashboard of each student's mastery and whole-class trends, privacy by design: data serves learning support and lecture tuning — not surveillance.

THE BREAKTHROUGH

For the first time — faculty see every single student's mastery in their course.

WHERE STUDENTS ARE STUCK

The AI analyzes questions, errors, response times.

WHICH CONCEPTS ARE MISUNDERSTOOD

Tracked per individual student — not class averages.

WHO EXCELS, WHO FALLS BEHIND

Progress over time, intervention in time.

→ A class of 200 — cared for like 1-on-1.

Privacy by design: data serves learning support, not surveillance.

WHAT FACULTY GET BACK

5–10 HOURS SAVED / WEEK

The AI absorbs repeat questions, drafts quizzes, summarizes lectures.

DEEPER CLASS INSIGHT

Class-level insight: common questions, shared knowledge gaps.

PROFESSIONAL GROWTH

Time returned to research and one-on-one mentoring.

"AI takes the repetitive work — faculty keep the core work: guiding each student with insight they've never had before."

The All-in-One model

Your university provides the course materials — we handle everything else.

Five integrated components, operated and supported end-to-end — no in-house tech team required to start.

01 Sovereign cloud infrastructure

Runs on Mobifone Cloud, physically hosted in Vietnam.

02 AI from the world's leading labs

Anthropic, Alibaba, and other frontier providers — continuously updated to the latest models.

03 The AI Agent for Education platform

A dedicated pedagogy product layer with 5 core capabilities.

04 Knowledge base packaging

Turning raw materials into a queryable AI asset.

05 Operations & full support

Technical, faculty, and student support — with content-quality monitoring.

For university leadership

Spend 45 minutes watching an AI learn your own course.

A private demo for your leadership and deans: we ingest one chapter of your actual textbook and show live how the AI teaches from it.

A demo on your own course materials — no canned script
Discuss the pilot roadmap, operating model, and pioneer-school incentives
Answers on data sovereignty, security, and integration with existing systems

Request a private demo

We respond within 24 business hours.

Request received

The AI Lab for Education team will contact your institution within 24 business hours to schedule the demo.