You have spent two hours on a contract interpretation question. Eight considerations, three clauses that interact in odd ways, case law that cuts both ways, commercial factors. You know the answer is in there. But when you try to explain it, it comes out chronological and tangled: “well, first there’s this clause, and then there’s this other one, but actually there’s also…”
Your reasoning is sound. Your structure is the problem. And a model is not just for generating text — it is good at imposing structure on thinking you have already done.
The Pyramid Principle, briefly
Barbara Minto’s framework, which we leaned on for ten years building legal visualisations at Patroon, is simple: lead with your conclusion, then organise the support in a hierarchy. Top: the answer. Second level: the three or four main reasons. Third level: the evidence under each reason. Clarity comes from the logical dependencies between ideas, not from listing them in the order they occurred to you.
Most lawyers know the structure exists. The pressure point is using it while you are still working a problem out, because the structure forces three things: commit to an actual conclusion instead of “it depends”, group your reasoning instead of scattering it, and see whether the support actually holds the conclusion up.
The prompt
Organise my legal analysis using the Minto Pyramid Principle.
Here is everything — the question, the relevant clauses or provisions,
my analysis so far, the competing considerations, what I'm unsure of:
[dump it all here]
Structure it as a pyramid:
CONCLUSION — the answer to the question. Be specific. State
qualifications clearly, but don't hedge.
MAIN REASONS — the 3-4 reasons that support it. Each should be
independently sufficient or collectively necessary.
DETAILED REASONING — for each reason, the specific evidence, principles,
or logical steps beneath it.
WEAKNESSES — where is this vulnerable? What's the strongest
counterargument? What fact or rule would change the conclusion?
Be honest about gaps. I need to see where my reasoning is actually weak.The first output organises your thinking — and exposes it. You will immediately see that a supporting point does not actually support its conclusion, that two of your reasons are the same point wearing different clothes, or that your conclusion is stronger than the reasoning earns. That is the entire value. Then you keep going: “the reasoning under reason two feels thin — what am I missing?” You are not asking the model to solve the problem. You are asking it to mirror your logic back so you can sharpen it.
Where it changes the answer
A distributor wants to sell on online marketplaces. The agreement allows “retail and wholesale channels” but also requires the distributor to “maintain brand standards.” The manufacturer says marketplaces breach the deal. Can the client do it?
Run the pyramid and the first conclusion comes back as “likely yes” — but the weaknesses section shows real termination risk. Those two do not line up, and you can see it. So you push: is the brand-standards point a weakness in my legal reasoning, or a commercial risk? The distinction reorganises everything. Legally the contract permits marketplace sales; the language is broad and there is no explicit ban. The risk is commercial: the manufacturer might terminate anyway, even on weak grounds. The revised conclusion is sharper than where you started — the client has the contractual right, but should propose marketplace standards first and use the strong legal position as leverage rather than just proceeding.
You would not have reached that from your initial pile of thoughts. The structure forced you to separate the legal question from the business one, and the real answer was sitting in that gap.
This is the same logic Dutch courts follow in the civiel beslismodel — work through the layers of a dispute in order, and the structure itself surfaces what a single pass would miss.