Guide 01
How to Use an AI Whiteboard Effectively
An AI whiteboard works best as a visible reasoning surface, not as a place to paste a finished question.
Start with the problem
Write the objective, known facts, and unknowns before adding detail. A simple boundary around the current problem prevents a large board from becoming a collection of unrelated notes.
Build a questionable model
Use labels, arrows, equations, or blocks to show relationships. Then ask about one visible uncertainty and compare the answer against the model.
Iterate deliberately
Revise the drawing after each useful explanation. Finish with a short note in your own words so the board records understanding, not only conversation.
Mistakes to avoid
- —Waiting until the entire diagram is perfect before asking.
- —Putting several unrelated problems on the same working area.
- —Accepting an explanation without checking it against the canvas.
Key ideas
- •Define the problem before drawing detail.
- •Make relationships visible with labels and arrows.
- •Use follow-ups to test understanding.
Guide 02
How to Solve Problems Visually
Visual problem solving turns an abstract question into an inspectable model.
Externalise the system
Place components, variables, or events outside your head. Separate known information from guesses and label the direction of important relationships.
Choose the right representation
Use a flowchart for sequence, a block diagram for components, a graph for change, and a sketch for spatial relationships. Do not draw a diagram when a short list is clearer.
Revise the model
Test assumptions by changing one part at a time. If the model cannot explain an observation, update the model rather than forcing the answer.
Mistakes to avoid
- —Using decorative shapes instead of meaningful labels.
- —Hiding assumptions inside vague arrows.
- —Treating the first diagram as final.
Key ideas
- •A diagram is a model, not a picture.
- •Representation should match the question.
- •Revision is evidence of learning.
Guide 03
How to Study With an AI Tutor
Use an AI tutor to examine reasoning, not to skip it.
Attempt first
Put your own first attempt on the canvas. It gives you a concrete place to identify what you understand and where the reasoning breaks.
Ask for explanation
Ask why a rule applies, what a term represents, or why a step follows. Request a second explanation if the first one uses unfamiliar language.
Test yourself
Cover the answer, reproduce the reasoning, and explain the idea back in your own words. This turns guidance into durable learning.
Mistakes to avoid
- —Asking only for a final answer.
- —Copying steps without identifying their purpose.
- —Failing to check units, signs, or assumptions.
Key ideas
- •Your attempt is useful context.
- •Why questions reveal reasoning.
- •Independent re-solving checks understanding.
Guide 04
How to Learn Mathematics With a Whiteboard
A whiteboard makes intermediate mathematical reasoning visible and easier to audit.
Write one transformation per line
Keep algebraic changes separated so a sign, factor, or substitution can be checked. Annotate why a change is valid when the rule is easy to forget.
Draw the relationship
Use sketches for functions, geometry, vectors, trigonometry, or rate-of-change questions. A sketch can reveal domain, direction, or symmetry before calculation.
Check intermediate results
Estimate magnitude, substitute simple values, and compare units or graph behaviour. Verification belongs between steps, not only at the end.
Mistakes to avoid
- —Combining several algebraic operations in one jump.
- —Using a graph without labelling axes.
- —Treating a diagram as exact when it is only a sketch.
Key ideas
- •Readable working exposes errors.
- •Sketches support algebraic reasoning.
- •Check each transition.
Guide 05
How to Learn Physics With Diagrams
Physics diagrams help connect mathematical symbols to a physical situation.
Model the situation first
Choose the object or system boundary, then draw forces, motion, fields, rays, or circuit elements relevant to the question.
Label variables and units
Place directions, reference axes, and units near the diagram. This prevents equations from becoming detached from what they represent.
Check the model against reality
Ask whether every arrow has a cause, whether directions are consistent, and whether the diagram would change if a condition changed.
Mistakes to avoid
- —Mixing force and motion arrows without labels.
- —Using a circuit drawing as if it were a physical layout.
- —Skipping the system boundary.
Key ideas
- •Choose the system.
- •Label direction and units.
- •Connect equations to physical meaning.
Guide 06
Visual Problem Solving for Engineers
Engineering work benefits from a visible model of components, constraints, and interfaces before a solution is selected.
Decompose the system
Draw inputs, outputs, components, and dependencies. Mark interfaces where information, energy, material, or control crosses a boundary.
Record constraints and assumptions
Place constraints near the component they affect. Separate measured facts from assumptions so they can be revisited during review.
Use AI as a reasoning partner
Ask for explanations of a relationship or an alternative framing, then validate against specifications, calculations, experiments, and professional judgement.
Mistakes to avoid
- —Treating a block diagram as a complete design.
- —Leaving failure modes implicit.
- —Using AI output without independent validation.
Key ideas
- •Draw interfaces and dependencies.
- •Make constraints explicit.
- •Validate every important conclusion.
Guide 07
How to Ask Better Questions to AI
A strong question identifies the relevant context, the exact uncertainty, and the kind of answer that would help.
Give focused context
Instead of “Explain this,” ask “In this labelled circuit, why does current stop when this switch is open?” The board can hold the setup; your question should name the target.
State the desired depth
Ask for a simple explanation, a derivation, an example, a check of assumptions, or an alternative approach. Depth is not the same as length.
Use follow-ups to refine
Ask what changed, what assumption was used, or where your reasoning diverged. Follow-ups are often better than one oversized prompt.
Mistakes to avoid
- —Asking multiple unrelated questions at once.
- —Leaving the target object unnamed.
- —Treating a confident response as proof.
Key ideas
- •Name the uncertainty.
- •Ask for the format you need.
- •Use AI answers as hypotheses to check.
Guide 08
How to Use Voice AI for Problem Solving
Voice is useful when a spoken question keeps you closer to the drawing than a typed prompt would.
Think aloud in short turns
Describe the immediate issue, then pause to inspect the response against the board. Short turns make it easier to redirect the discussion.
Keep visual reference
Use voice while looking at the relevant diagram or calculation. Name the labelled item or step you are discussing.
Know when to type
Switch to text for exact equations, code, formal constraints, or content you want to edit carefully before sending.
Mistakes to avoid
- —Trying to dictate a complex equation without showing it.
- —Using voice for several unrelated topics at once.
- —Not correcting a mistaken interpretation early.
Key ideas
- •Voice supports conversational exploration.
- •The canvas anchors the discussion.
- •Text remains best for exact notation.
Guide 09
How to Break a Complex Problem Into Smaller Problems
Decomposition creates smaller questions that can be checked before they are recombined.
Map objective, knowns, and constraints
Write the final objective first. Then distinguish information you have from information you need and conditions you cannot change.
Divide by components and dependencies
Draw subproblems as blocks and connect them with arrows. Solve independent pieces first, then identify the results that feed the next step.
Reconnect and validate
Bring results back to the original objective. Check whether the combined answer respects every constraint and assumption.
Mistakes to avoid
- —Dividing a problem without tracking dependencies.
- —Solving subproblems that do not affect the objective.
- —Forgetting to validate the recombined result.
Key ideas
- •Make the objective visible.
- •Track what depends on what.
- •Validate the whole system again.
Guide 10
How to Use Diagrams to Understand Complex Concepts
A diagram should make one relationship easier to see than it would be in prose.
Match the diagram to the information
Use concept maps for categories, flowcharts for process, timelines for sequence, and cause-and-effect diagrams for contributing factors.
Reduce visual noise
Use a small number of consistent shapes, labels, and arrow styles. If a line has no meaning, remove it.
Read the diagram actively
Ask what each connection means, what is missing, and which relationship would change under a new condition.
Mistakes to avoid
- —Using one diagram type for every problem.
- —Adding labels that repeat rather than clarify.
- —Ignoring direction and scale conventions.
Key ideas
- •Form follows question.
- •Less clutter improves reasoning.
- •Diagrams invite questions.
Guide 11
How to Learn From Mistakes Instead of Just Correcting Them
A corrected answer is less useful than knowing where and why reasoning first diverged.
Find the first divergence
Compare your steps with a sound method and locate the earliest incorrect assumption, definition, or transformation.
Classify the error
Distinguish calculation slips from conceptual errors. A sign error and a mistaken model need different fixes.
Re-solve independently
After understanding the correction, start again without looking at the answer. Record the check that would catch the same error next time.
Mistakes to avoid
- —Only looking at the final line.
- —Calling every mistake careless.
- —Memorising a correction without rebuilding reasoning.
Key ideas
- •Find the first wrong step.
- •Name the error type.
- •Practice the corrected method.
Guide 12
A Complete Visual Learning Workflow
Visual learning becomes reliable when drawing, explanation, revision, verification, and summary form a repeatable loop.
Draw and explain
Start with a rough model, then ask about the relationship you cannot yet explain. Keep the question narrow enough to test the answer.
Question and revise
Modify labels, arrows, or equations as you learn. A revised board is evidence that the model has improved.
Verify and summarise
Check the result with an example, unit, boundary case, or independent derivation. End with a concise summary you can recreate later.
Mistakes to avoid
- —Treating explanation as the finish line.
- —Leaving old assumptions on the canvas.
- —Skipping an independent summary.
Key ideas
- •Problem → draw → question → revise.
- •Verification protects against plausible errors.
- •Summaries turn a session into reusable knowledge.
Related resources