THINK BEYOND ANSWERS

Understand it.
Make it yours.

Turn AI explanations into knowledge you can use—with a four-part grid you write, check and connect.

Write your first grid See an example

AI made it sound clear. Can you explain it when the question changes?

METACOGNITIVE TRAINING

It feels clear. Can you check it?

Predict your confidence. Explain independently. Use evidence to choose your next step.

Start a training session

01 / Something you can keep building

A notebook that grows
with your understanding.

Write the relationship on the left.
See the connections on the right.

Your grid notebook

Maths example · 3 entries

Select an entry to trace its connection and check status.

Example learning records. Select the starting point to highlight its connection in the map.
Relationship	Conditions	Result
supports	The outside factor multiplies every term inside the brackets.	Expand bracketsChecked
helps	An expression contains brackets that need to be expanded.	Simplify expressionsReview in 3 days
helps	Each transformation preserves the equation’s solutions.	Solve linear equationsNot checked
support	Add or subtract the same quantity on both sides; multiply or divide by the same non-zero quantity.	Solve linear equationsNot checked

Interactive example. This does not change your saved draft.

Your understanding map

Trace each connection back to an entry

CheckedReview due / not checked

CheckedExample check status

Distributive lawsupportsExpand brackets

ConditionsThe outside factor multiplies every term inside the brackets.

Return to this notebook entry

An illustrative product concept. Each connection has a recorded relationship and conditions. Review reminders and automatic connections are planned features.

Write your first grid

02 / From an explanation to something you can use

Write it. Check it. Connect it.

AI can help. You still do the thinking.

01
Write one relationship.
Start with a real question. State what you know, how the ideas connect, when the relationship holds and what follows.
02
Try a different question.
Explain it in your own words. Test it in a new situation. Record what needs checking or revisiting.
03
Build on what stays.
Connect the entry to earlier ideas. Revisit it later, so your notebook records more than a moment of recognition.

Try it with one question

03 / Start where you get stuck

Six familiar problems.
A clearer way through.

You don’t need to learn the whole method first.
Choose a situation that feels familiar.

Try the Card Lab: add a context or condition and watch the result change →

01 / History

I know the Industrial Revolution timeline. I still can’t explain what changed.

Connect changes in production to changes in how people lived.

See how entries connect

CStarting point

What do you start with?

Factory jobs

rRelationship

How does it connect?

draw workers towards

@DConditions

When does it hold?

Where factory employment grows in towns. Transport, housing and local conditions also matter.

C′Result

What follows?

Growing towns and cities

02 / RAG interviews

The right document appears in the logs. Why is the final answer still wrong?

Finding the document is one step. Check what the model actually receives.

See how entries connect

CStarting point

What do you start with?

A retrieved document

rRelationship

How does it connect?

can support

@DConditions

When does it hold?

The relevant passage reaches the final context, and the answer uses its evidence accurately.

C′Result

What follows?

An evidence-grounded answer

03 / TopK

I said “use a min-heap.” Then the interviewer asked, “What if K is very large?”

Explain when a method fits before calling it the best choice.

See how entries connect

CStarting point

What do you start with?

A size-K min-heap

rRelationship

How does it connect?

retains

@DConditions

When does it hold?

Useful for a stream or relatively small K. For static data, compare selection and sorting approaches.

C′Result

What follows?

The largest K items

04 / Maths

We’ve practised expanding brackets. A new question still turns −2(x − 1) into −2x − 2.

Check whether the outside factor reaches every term, including the negative one.

See how entries connect

CStarting point

What do you start with?

−2(x − 1)

rRelationship

How does it connect?

expands to

@DConditions

When does it hold?

Multiply −2 by both x and −1.

C′Result

What follows?

−2x + 2

05 / Business

I can define renewal rate. But I freeze when asked why the business depends on it.

Connect the metric to the economics it is meant to explain.

See how entries connect

CStarting point

What do you start with?

Customer acquisition cost

rRelationship

How does it connect?

is recovered through

@DConditions

When does it hold?

In a subscription business where the first period’s gross profit does not cover acquisition cost.

C′Result

What follows?

Gross profit from continued renewals

06 / Research

My supervisor said “p < 0.05 doesn’t mean your conclusion is right.” I’m lost.

Put the assumptions back into the definition before interpreting the number.

See how entries connect

CStarting point

What do you start with?

A p-value

rRelationship

How does it connect?

measures

@DConditions

When does it hold?

Assume the null hypothesis and statistical model hold; define the relevant tail or extremeness criterion.

C′Result

What follows?

The probability of an equally or more extreme test statistic

What are you trying to understand?

Put it into a grid

Illustrative examples of the method. They are not automatically generated diagnoses or claims about a learner.

Use AI with judgement

Make the reasoning explicit.
Give AI better direction.

When you can state the relationship and its conditions, you can question an answer and explain the judgement behind it.

Build understanding you can use

Go beyond recognising an answer.
Explain why it applies.

A new example tests whether the relationship still makes sense when the surface details change.

04 / Research informs the method

Learning science,
made writable.

One grid brings together ideas from more than twenty learning-science studies.

Full bibliography (Chinese)

Learning research connected to the four grid fields
In the grid	Related research
Starting pointWhat you already know	Prior knowledge and learning hierarchiesAusubel · Gagné
RelationshipHow ideas connect	Self-explanation, elaborative questioning and concept mappingChi · King · Novak
ConditionsWhere a relationship holds	Conceptual change and the limits of an existing explanationPosner
ResultWhat follows, then test it	Generation, retrieval practice and spaced practiceSlamecka & Graf · Roediger & Karpicke · Cepeda

The research helps explain the design: activate prior knowledge, explain relationships, state their limits, then retrieve and test what you have learned.

Evidence for these learning mechanisms is not the same as direct validation of MindLattice as a product. We do not promise grades, admissions or job offers.

Developed through
AI and education research.

Methodology

Chao Li

OCG methodology author and AEGT principal researcher, working on AI education and AI psychology.

Education collaboration

Research meets practice

Co-developed with an education researcher with a doctorate from the University of Otago and experience in educational practice.

Team details (Chinese)

05 / Different ways to keep building

From your notebook
to a shared body of knowledge.

Choose the support your work calls for.

Individuals01

Your grid notebook

Keep building on what you learn.

AI-guided grid writing
Check questions and review prompts
A record of your understanding

Optional: a course in building your own grids.

Monthly or annual subscription planned

In development. Pricing to be announced.Explore the method

Schools & teams02

A notebook for every learner

See individual progress and shared sticking points.

A notebook for each learner
Class or team overview
Shared learning progress

Optional: teacher or facilitator training.

Annual per-learner pricing planned

In development. Pricing to be announced.Explore the method

Organisations03

Expert knowledge licensing

Make expert reasoning usable in your own AI system.

Expert-reviewed domain structures
Relationships with explicit conditions
Updates as the domain structure develops

Optional: a custom domain map.

Annual licensing planned

In development. Pricing to be announced.Explore the method

Product plans are in development. Subscriptions, team dashboards, AI-guided writing and licences are not available to purchase here.

Start with one question

Write your first grid.

You can try the method now. This is a manual worksheet: write your explanation, add a question that would test it, then save or download your draft.

Try the worked example

We expanded −2(x − 1) to −2x + 2. Now change the numbers.

Expand −3(2x − 4).

This check covers one algebra example. It does not automatically assess your own topic or prove lasting mastery.

Understand it.Make it yours.

AI answers come and go.Make the thinking your own.

It feels clear. Can you check it?

A notebook that growswith your understanding.

Your grid notebook

Your understanding map

Write it. Check it. Connect it.

Write one relationship.

Try a different question.

Build on what stays.

Six familiar problems.A clearer way through.

Connect changes in production to changes in how people lived.

Finding the document is one step. Check what the model actually receives.

Explain when a method fits before calling it the best choice.

Check whether the outside factor reaches every term, including the negative one.

Connect the metric to the economics it is meant to explain.

Put the assumptions back into the definition before interpreting the number.

Make the reasoning explicit.Give AI better direction.

Go beyond recognising an answer.Explain why it applies.

Learning science,made writable.

Chao Li

Research meets practice

From your notebookto a shared body of knowledge.

Your grid notebook

A notebook for every learner

Expert knowledge licensing

Write your first grid.

Try the worked example

Understand it.
Make it yours.

A notebook that grows
with your understanding.

Six familiar problems.
A clearer way through.

Make the reasoning explicit.
Give AI better direction.

Go beyond recognising an answer.
Explain why it applies.

Learning science,
made writable.

From your notebook
to a shared body of knowledge.