The personalized teaching layer for AI

It doesn't just answer.

It learns how you learn — and how to teach you next.

General models solved knowing the answer. Aptiz AI builds the personalized teaching layer on top — it works out who the student is, where they're stuck, and what to teach next.

See the live demo For investors →
What a real teacher needs Sheet 01
01Knowledge
Knowing the answer
Solved by LLMs
02Teaching adaptation
How to teach this student
Our layer
03Learning state
What to teach right now
Our layer

01 · The problem


General AI waits for a good question — then stops at one answer.

Not the bottleneck

Generating knowledge

The real bottleneck

Deciding how to teach it

02 · What makes us different


Ordinary AI tutor Aptiz AI
Waits for the student to ask preciselyInfers the next step from work, errors & behavior
Generates one explanationGenerates several teaching moves and ranks them
Personalization stops at tone & examplesPersonalizes the move, order, difficulty & hints
Chat ends = doneMust be verified by a new, delayed task
Logs the conversationMaintains a correctable student state
Optimizes satisfaction / time-on-appOptimizes mastery, transfer, memory, confidence

03 · Live demo — try it

Illustrative

Same student, same error. Pick a teaching move — or let the system choose — then see why, verify it, and watch the model update. Switch students to see the same error get a different move.

Student Mia · Year 9
Goal Solve linear equations with fractions
Error detected Multiplied only one side by the denominator

Three candidate teaching moves

04 · Why we can optimize teaching


Recording every learning trace is where AI beats any human — so we spend all our effort on teaching well.

A pile of chat logs isn't an advantage. Structured cause-and-effect — state → move → result → mastery — is. It compounds across students, subjects and regions: the more it's used, the better it gets.

LayerWho's strongerOur focus
KnowledgeLLMsNo
Recording & trackingAI, by natureFoundation
Targeted teachingBuilt by data flywheelCore

05 · What counts as learning here


Step 1

Initial error

Step 2

Intervention

Step 3

Mastery on a new task

We accept one definition of success: unprompted transfer, still remembered later.

Not evidence of learning:

watch time cards completed clicks "the kid likes it"

06 · Vision


For the first time, scale and personalization can happen together.

Scaling education used to mean standardizing it. AI breaks that trade-off — which reorganizes not a software category, but how every student understands, practices, gets feedback, and plans what's next.

i · Who decides where learning goes

Bus

Traditional school

Fixed route, time and group pace. Fall off the group's speed and you're left behind.

Carpool

Human tutor

More flexible, but bound by price, time and teacher supply — never everyone's infrastructure.

Where Aptiz AI takes it

Private car

Personal AI tutor

The learner sets the destination and leaves anytime; the system adapts the route to them.

ii · How much the system drives — our roadmap

Phase 1 · Manual

Today's general AI

Powerful — but the student must judge what they don't know and what to do next.

Phase 2 · Automatic

Decides within a frame

The system chooses inside a human-designed teaching framework.

Phase 3 · Self-driving

Finds new methods

The system discovers teaching combinations no human wrote down.

iii · Equity = the market-expansion flywheel

Not a slogan bolted to the end — it comes straight from the cost structure. Lower marginal cost makes people who could never afford a tutor into users. Equity means the whole market gets bigger.

Cost falls

More students reachable

More learning evidence

Better adaptation

…and the loop repeats — each turn makes teaching better and cheaper, pulling in more students and expanding the market again.

iv · The market, bottom-up

Entry

Paid personalized help

High-intent, well-defined tutoring & exam prep.

Expansion

A subscription product

High-frequency practice, explanation & planning.

Endgame

Every student's learning layer

The core interaction layer of next-gen education.

Mission

The help a child gets shouldn't depend on income, postcode, or luck with a teacher.

We don't promise to erase every gap. We aim to lower how much high-quality, personalized teaching depends on family income, region and teacher supply.

Early customers

Work with us

A limited Founding Pilot — a personalized AI learning agent with human quality calibration.

Explore the pilot →

Investors

Let's talk

Deck, pilot data, unit economics and milestones on request.

Investor contact →

07 · Investors


Leave your details — we'll message you.

Deck, pilot data, unit economics and milestones on request. Tell us how you prefer to be reached and we'll come to you there, usually within two business days.

Prefer email? Write to [email protected] directly.

We never share your details with third parties.