How a lesson works

Find the gap. Build the idea. Prove it clicked.

Manthika turns one topic into a short, adaptive climb. It starts where the learner is, uses their own AI for small investigations, and only calls an idea mastered when they can use it for themselves.

In the lesson itself, that is just three steps: Find your level · Practice · Complete.

a topic
Checks where you standa few quick questions, before any teaching
Teaches the one idea you're missingnot the next item in a fixed list
Asks a real questionauto-graded on the spot
clicked?not yet — it comes at it another wayyesyesthe next idea
not yet — it comes at it another way
The loop a good teacher runs by hand: find the gap, teach it, check it honestly — and never move on until it clicks.
01 · Place

Start at the right rung

No account or setup. The learner picks their name and answers four quick questions. Each angle is marked solid, shaky, or not yet, so the lesson can skip what they know and begin at the first real gap.

“An app guesses a house’s price from its size. Is the output a number, a group, or a picture?”

Why we start here, not at a fixed lesson one →

A placement question on the live lesson, with one question and three answers on the phone screen.
02 · Climb

Build one idea from five angles

The app serves one card at a time and always works on the lowest shaky angle. If an advanced question exposes a weak foundation, the path quietly drops back and repairs it.

Why five angles, and what a right answer actually proves →

Clarity
What is it?
Explanation
How does it work?
Background
Why is it needed?
Implications
Where can it fail?
Connections
What else does it affect?
75 · next topic90 · mastery
ClarityExplanationBackgroundImplicationsConnections
Clarity
Explanation
Background
75 — next topic unlocked
Implications
Connections
90 — full mastery
The first three solid clears 75 — next topic unlocked. All five, applied to something new, clears 90 — full mastery.
Clarity
Explanation← picks here
Background
Implications
Connections
It always picks the lowest open rung — never an advanced one while an earlier rung still needs work.
03 · Investigate

Use AI, then think past it

Some cards send the learner to the AI their family already uses; others provide seed cases inside the lesson. Either way, they bring examples back and answer follow-ups about their own work.

“Ask your AI: Explain how regression guesses a new value, like I’m 12.”

Because the next question refers to their examples, copying a generic chatbot answer is not enough. The AI conversation stays outside Manthika. Pasted notes stay on the phone unless the learner signs in; then they sync with the rest of their progress.

An investigation screen where a learner selects seed cases and explains the pattern in their own words.
An ask-your-AI nudge shown inside a live question card.
The appgives a short research task, tied to the lesson
📋 prompt
Your own AIChatGPT, Gemini, Claude — whatever your family already hasthe app never sees this
your paste
The appasks a follow-up about YOUR paste
The question after is about yourpaste, not a generic fact — so copying it into a chatbot won’t get you the answer.
The progress screen showing the 75 floor cleared and the remaining climb toward full mastery at 90.
04 · Prove

Two useful finish lines

At 75, the three core angles are solid and the next topic unlocks. At 90, all five are solid and the learner has applied the idea to a new case. Facts let them move on; transfer proves mastery.

If the same gap stays stuck, the app pauses kindly, saves progress, and invites the learner to reread the exact book page before returning.
Ravi
🏆
Meena
🌱
ClarityBackgroundConnections
Same five angles, two different climbs. Meena's pause is real — the figure doesn't draw a recovery the page doesn't claim.

A confident learner can move quickly to 90. A struggling learner gets a shorter path, retries at the foundation, and stops before frustration takes over.

One chapter

All the forms it has today

No chapter exists yet in every form the factory can make. This is the one that has reached the most: the same Class 10 Mathematics chapter, written once, shown below in each form it has reached so far.

1. Book page — Class 10 Mathematics, Chapter 1 (Real Numbers), as it exists today.
1. Book page
2. Open web chapter — Class 10 Mathematics, Chapter 1 (Real Numbers), as it exists today.
2. Open web chapter
3. Course screen — Class 10 Mathematics, Chapter 1 (Real Numbers), as it exists today.
3. Course screen
4. Feed card — Class 10 Mathematics, Chapter 1 (Real Numbers), as it exists today.
4. Feed card
5. Worksheet — Class 10 Mathematics, Chapter 1 (Real Numbers), as it exists today.
5. Worksheet
6. Revision sheet — Class 10 Mathematics, Chapter 1 (Real Numbers), as it exists today.
6. Revision sheet
7. Sample paper — Class 10 Mathematics, Chapter 1 (Real Numbers), as it exists today.
7. Sample paper
8. Flashcard — Class 10 Mathematics, Chapter 1 (Real Numbers), as it exists today.
8. Flashcard

8 forms measured on 1 October 2026, from one Class 10 Mathematics chapter — the chapter that has reached the most forms today, not a typical one.

Across a whole subject

What a belt means

A belt sums up how far a learner has gone across an entire subject — not one topic, one quiz, or one score. Each subject keeps its own belt, and only the subject's core topics count toward it.

YellowSAYSThe subject is underway.EARNSTouch any core topic — answerone question in it.GreenSAYSHalf the subject's core topicsare past the first step.EARNSGet half the core topics pastthe first step.BrownSAYSEvery core topic is past thefirst step, and at least oneis fully mastered.EARNSClear the first step in everycore topic, and fully masterat least one.BlackSAYSAlmost every core topic in thesubject is fully mastered.EARNSFully master at least nine inten core topics.

Only the subject's core topics count toward its belt — enrichment topics don't move it either way.

A belt never turns into a mark or a percentage. It stays a belt.

Where it is today

One complete lesson, end to end

● Working now

Grade 7 Regression, the full adaptive path, own-AI investigations, local learner profiles, saved progress, progress that follows a signed-in learner to another device, real book-page links, and one topic (Adult Numeracy: Fractions to Algebra).

○ Next, not hidden

More topics, more languages and richer support for open answers are still to be built.

How a chapter is made

This chapter’s finished forms are counted just below.

A subject is written once, drawn once, and checked before anything ships. From that single checked source, the factory re-expresses it into every form a line needs.

Why the loop works

Designed around understanding, not screen time

The ‘Who’s learning?’ screen on a shared phone, showing two local learners already saved on the device and a field to add a third, with no sign-up or account required.

Adaptive from the first minute

A short placement check finds the lowest shaky idea. Strong learners skip ahead; stuck learners step back.

AI is a tool, not the teacher

Learners use the AI their family already has, then return to think with the examples they found.

Mastery requires transfer

Remembering facts can unlock the next topic. Full mastery needs applying the idea somewhere new.

Made for a shared phone

No login to start, one question per screen, several learners on one phone, and progress saved on that device.

The quickest way to understand the loop is to take the first question.

Try the live lesson →