Think
Classic thinking models, carried out on your problem
You know what a red team is and you have heard of a steelman. The hard part was never knowing the list — it is remembering the right method at the moment you need it, and then doing the work of applying it. Pick one here and it comes back already done, on your problem and your material.
Carried out, not named
A model that only tells you its own name is a flashcard. What comes back here is the method, done.
- A red team comes back as the specific ways your plan fails, from the adversary’s side. An assumption audit digs out what the argument takes for granted and questions each one.
- First principles strips away the analogy and the convention and reasons up again from what is left. A latticework runs the same problem through models from several disciplines and shows where they disagree.
- Point the card at documents in your folder and the method is applied to their actual text — a steelman of the paper you are arguing with, an evidence-strength table over the reports you were sent.
Aim it at one line, or at several cards at once
A method is only as sharp as what you point it at. The target can be much smaller than a card — or much bigger.
- Every line on a card has its own skill button. Run a causal chain on the one point you do not understand, and the rest of the card stays as it was.
- Select several cards, or drop one card onto another, and a critique, structure or converge model runs across them as one set. Three option cards become one option matrix; four sources become one map of where they agree and where they split.
- The menu only lists what fits what you picked. Comparative analysis and consensus-and-disagreement wait until there are several cards to compare; the image skills appear only on an image card.
One hard decision, three models deep
Write the problem as a card
Specific enough to be wrong about: not “our pricing”, but “should we drop the monthly plan before the spring launch?”. Attach the notes, the reports, the thread with the numbers — the model is only as grounded as what the card points at.
Give each option its own card
Keep the monthly plan, drop it, or drop it only for new customers. Three cards, side by side.
Put all three on one ruler
Select them and apply an option matrix (a decision matrix): every option scored against the same criteria, in one table.
Audit the line that is winning
Run an assumption audit on the one row the matrix favours. This is where most decisions turn out to rest on one premise nobody checked.
Converge and hand it on
Trade-off analysis, then an action plan: what to do now, what next, and what waits for which condition. Save it as a note, or send it to the writing pipeline with the methods attached.
Other problems people bring to it
The mechanism is the same every time — a model, applied to a card, producing branches you judge. What changes is which models, in what order, and what you walk out holding.
An idea that is still soft
Brainstorm it wide, run SCAMPER over the version you like, cross-breed the parts that survived, then distil. You walk out with a shortlist you picked yourself, not a list of twenty options.
A subject you want to actually understand
Break the question into parts that do not overlap, strip it back to first principles, trace the causes and effects, and map how the terms relate. What comes out is a structure you could explain without the notes.
A field you are learning from scratch
Check what you are missing underneath first, get a study order, then have it test you and tell you which mistakes cluster where.
An argument you need to test
Lay out its argument, dig out the assumptions it never states, build its strongest opponent, then attack your own reading of it. You end up with a position, not a summary.
Written up end to endTwo of these, with the models named in order and an honest note on where they get hard: learning a new field and thinking through a decision.
Each method keeps its own shape
Some methods are not a list of branches. A comparison needs columns; a quadrant needs two axes; a history needs an order.
- Methods with a fixed form keep it. Six thinking hats always comes back as six; SCAMPER as seven; a steelman builds the other side at its strongest before anything else happens.
- An option matrix, a comparative analysis, an evidence-strength check and a prioritisation come back as one table: the options down the side, shared criteria across the top, every cell readable against its row and column. SWOT comes back as a four-way grid.
- A causal chain and a timeline come back in order, as a strip, rather than fanned out as if the order did not matter. A table can be split into a cluster of cards to work on one row, and gathered back afterwards — nothing is lost in either direction.
A group that asks you instead of answering
Reading an explanation feels like understanding. Being asked is how you find out whether it was.
- Self-quiz sets questions on the card and marks the ones with definite answers on the spot. The report is by what you got wrong, not by score — and it flags the ones you said you were sure about, which are the ones worth knowing.
- Feynman explanation has you say it in plain words and shows where you got stuck. Prerequisite check has you rate yourself on what the topic assumes, so the gap is found before it trips you.
- Bloom’s ladder shows which level you have reached on a topic — remembering, understanding, applying and up — and what the next level would ask of you.
The method goes with it into the write-up
An analysis that cannot say how it was reached is an opinion. When a branch goes to the writing pipeline, the models you ran on it go too.
- You choose how much of the method shows: conclusions only, angles explained without naming the tools, or methods fully shown — the framework named, with notes on how it was applied and where it falls short.
- Reports and papers default to showing the methods, with a “Framework & methods” section of their own. Articles and books default to weaving the angles into the argument instead.
- The branch arrives as the outline, so the structure you judged on the canvas is the structure of the draft.
The full menu: forty-four built-in skills
Most of them are named classic methods; a few are plain moves like brainstorming or pulling out key takeaways. They are grouped by what the problem needs, not alphabetically — so the question you ask the menu is “what am I stuck on?”, not “what was that one called?”.
Diverge
Brainstorm · Analogical transfer · Counterfactual · SCAMPER · Six thinking hats · Morphological combination. For when the problem is still soft and every idea so far is the obvious one.
Dig deeper
Decompose (MECE) · Causal chain · First principles · Clarify a concept · Latticework of mental models. For when you cannot yet say why something happens.
Critique
Red team · Assumption audit · Evidence strength · Steelman · Bias check · Socratic questioning. For when you already like your answer a little too much.
Structure
Book guide · Interpret · Concept map · Timeline · Learning path · SWOT · Work breakdown · Comparative analysis. For when you have the pieces but not how they fit.
Converge
Option matrix · Trade-off analysis · Prioritize · Key takeaways · Consensus and disagreement · Retrospective · Writing outline · Action plan. For turning a map into a decision.
Learn
Self-quiz · Feynman explanation · Prerequisite check · Bloom’s ladder · Common mistakes. The one group that asks you questions instead of answering them.
Not in the six groupsOn an image card, a separate Look at image group appears instead: read an image, chart to table, rebuild structure, design review, tidy handwriting and compare images — see images. A starter library adds ten more you install only if you want them: pre-mortem, Porter’s five forces, business model canvas, stakeholder map, system loops, Fermi estimate, user journey, minimum viable test, borrowed perspectives and deliberate practice design.
Where it gets hard
- A model is a lens, not a verdict
A red team finds holes; it does not tell you whether the plan is worth doing anyway. Every method here produces material for your judgement, and none of them replaces it.
- A vague problem gets a vague analysis
SWOT on “the business” produces four boxes of generalities. The card in the middle sets the ceiling for every model you put on it.
- The learning group makes you answer first
It asks before it tells, because being told is what made you think you understood. If what you wanted was the explanation, use Interpret or Clarify a concept instead.
- Your folder joins only when you switch it on
A board works on what you attached. That keeps a critique from being anchored to conclusions you already reached, but it also means an analysis is only as grounded as the material you pointed it at.
Common questions
Is there a tool that applies mental models to my problem?
That is what the skills on this canvas are. Each one is a classic thinking model written down as instructions — SCAMPER, six thinking hats, first principles, a latticework of mental models, a red team, an assumption audit, an option matrix, SWOT, Bloom’s ladder — and applying it to a card carries the method out on that problem, one branch per angle.
Which thinking models are included?
The canvas ships with forty-four built-in skills in six groups — diverge, dig deeper, critique, structure, converge and learn. Most are named classic methods; a few are plain moves such as brainstorming or pulling out key takeaways, and six only appear on image cards. Ten more can be installed from a starter library, including a pre-mortem, Porter’s five forces, a business model canvas, a stakeholder map and a Fermi estimate.
How do I know which model to use?
Start from what the problem needs rather than from the list. If it is still soft, diverge; if you do not understand why something happens, dig deeper; if you already like your answer too much, critique it; if you have to choose, converge. The menu is grouped that way, and it only lists the models that fit what you selected.
Can I apply a model to several cards at once?
Yes. Select several cards, or drop one card onto another, and a critique, structure or converge model runs across them as one set — three option cards become one option matrix, several sources become one map of where they agree and disagree. You can also aim a model at a single line inside a card instead of the whole card.
Can I add my own thinking model?
Yes. Skills are plain text. Copy a built-in one and change it, or describe a method you use and have a skill written for you. It then appears on the same menu.
Does it just name the model, or actually apply it?
It applies it. A red team comes back as the specific holes in your plan, a SWOT as a filled four-way grid with cross-quadrant moves, an option matrix as your options scored against shared criteria — on your problem and the documents you pointed it at.
See also
Option matrix, trade-offs, assumption audit and an action plan, end to end on a real choice.
The learning group end to end: what you are missing, a study order, and a test that tells you where you are wrong.
The canvas itself — starting points, branches, and how a map turns into an outline.
Book guide, Interpret and Concept map, run beside the document you are reading.
The same methods in chat and in the editor, and how to write one of your own.
Where an analysis goes once it is worth writing up — with its methods attached.
The definition on its own — and why knowing the list is not the same as using one.
The definition on its own — and where mind maps stop working.