In the legal field, the application of AI is far beyond imagination. This article will share 5 “king bomb” prompts to help lawyers turn AI into a “thinking plug-in”. From analogous analysis of case facts and adversarial review of contracts, to dismantling logic, sifting through hot information, and translating legal terminology, these skills make AI no longer a simple tool, but a powerful assistant to help lawyers expand their thinking and improve their professional capabilities.
Recently, the AI wind in the legal circle has made people panic.
Those crash courses on the market that “AI empowers law” are more energetic than those who work.
I looked at what they taught, what “help me write a contract” and “summarize the case”, to put it bluntly, it is to use AI as an intern who will make laws for you at any time.
This is called laziness, not empowerment.
The real way to play is to turn AI into your “plug-in”, a “super plug-in” that helps you expand your thinking, simulate confrontation, and open pictures in the information fog.
This requires you to understand its language, or in other words, you have to learn to speak “black words” to it.
Today I provide some useful prompt frameworks for lawyers.
These are not for you to copy and paste, you have to understand the logic behind it, and then fill it in according to your case.
B-end product manager’s ability model and learning improvement
The first challenge faced by B-end product managers is how to correctly analyze and diagnose business problems. This is also the most difficult part, product design knowledge is basically not helpful for this part of the work, if you want to do a good job in business analysis and diagnosis, you must have a solid …
View details >
Thoroughly, at least in the AI tool, you can get rid of the vast majority of peers around you who are still starting with “Hello, please help me…”.
01 Don’t let it look for the law, let it be an analogy of “case skeleton”
The most common mistake newbies make is to use AI as a legal library, and ask: “Are there any cases of virtual property inheritance?” “It turns out to be a pile for you.
The strength of AI is not accurate memory, but pattern recognition.
So, you have to ask a different question, not “find”, but “analogy”.
Try this prompt framework:
You now play the role of a senior civil and commercial judge, conducting rigorous analog case analysis. Please find and analyze the Guiding Cases or Gazette Cases in your knowledge base based on the ‘Fact Matrix’ of our cases below. The focus of the analysis is not on the similarity of the facts, but on the ‘reasoning part’ of the judgment, to see if its logical deduction process can be used as a reference for us.
The facts matrix of our case is as follows:
– Core facts: [For example, the parties have entered into a source code licensing agreement, but it is not clear whether the scope of the license includes derivative versions.] ]
– We claim: [For example, we claim that the other party’s actions constitute copyright infringement and infringe on the right of adaptation.] ]
– Core Disputes: [For example, whether the term ‘software’ in the agreement can be interpreted as a derivative version that extends to the future.] ]
Your analytics report must include:
1. Case matching degree ranking and similarity analysis.
2. Focus on analyzing the ‘chain of reasoning’ of the judgment in this case, and what inspiration or risks it has for us.
3. Based on the cases found, identify potential risks in our scenario that I am not aware of.
Look, what you give AI is not a keyword, but a “skeleton”.
You force it to understand the internal structure of your case, and then match the structure of other cases.
That “chain of reasoning” is the soul of the verdict, and this is what AI should do.
02 Don’t let it write contracts, let it be an “imaginary enemy” to conduct stress tests
“Help me write an equity transfer agreement”, this is the laziest and most dangerous usage.
The value of a lawyer lies in anticipating and allocating risk.
So, don’t let AI be your secretary, let it be your opponent, the opposing lawyer who tries his best to exploit loopholes in your contract.
Next time you review a contract, try this:
You’re now an extremely cunning top business lawyer, and your client is about to enter into this [contract type] agreement below with my client. Your only goal is to maximize the interests of your customers and lay the risk for my clients. Please review the following contract text one by one from the perspective of the most severe ‘adversarial review’.
[Paste the full text of your draft contract here]
Your review report should be like an internal memo that clearly states:
1. Ambiguity: What clauses are not clearly defined and can be used by you in the future?
2. Hidden traps and power asymmetry: Which clauses seem fair but actually make us passive? Where are the protections that are critical to your customers missing?
3. Ultimate attack strategy: If a dispute arises in the future, which clause will you choose to attack the agreement as a breakthrough? Explain your attack logic.
Please write in an extremely confident and slightly aggressive tone, with the intention of making me wary.
With this method, AI becomes your whetstone.
Every “trap” it points out is a “fortress” that you need to reinforce next.
After two rounds of tossing like this, the quality of your contract will be more than one notch higher.
03 Don’t let it summarize your views, let it be a “lever” to dismantle your logic
Litigation is a logical contest.
We write proxies for fear of falling into our own logical closed loop.
You need a 24-hour online “sparring” that is dedicated to lifting the bar with you.
After you’ve built a core argument, use this prompt to “attack yourself”:
I will present a legal argument in which you play the role of a professor of logic and a Socratic questioner. Your task is not to refute my conclusions, but to stress test my argumentation process itself with the strictest standards.
My argument is as follows:
– Premise (legal provisions): [State the legal provisions or doctrines you are based on.] ]
– Minor premise (factual claim): [State the facts you want the court to find.] ]
– Conclusion: [Your final claim.] ]
Now, please start your work and don’t give any solution, just pure questioning:
1. Attack premise: Are there exceptions to my premise? Are there other reasonable explanations for my factual claims?
2. Look for logical jumps: Is there something I take for granted in my reasoning?
3. Mock Judge’s ‘Soul Torture’: What three sharp questions is a skeptical judge most likely to ask me to challenge?
This process may be uncomfortable for you, but only after experiencing this “simulated disassembly” can your logic become airtight and prepare for all the questions at the trial in advance.
04 Don’t let it summarize the document, let it be an “intelligence officer” to filter “hot information”
When dealing with massive amounts of evidence, traditional keyword searches are too stupid.
AI understands semantics and context, and you can turn it into an intelligence analyst to help you identify “hot files” with specific “signals.”
You can give a command like this:
You’re a forensic accountant and intelligence analyst reviewing a batch of electronic evidence. Your task is not to summarize the content, but to act like a sieve, filtering and identifying files or fragments that meet the following ‘hot signals’ and extracting the original text.
[Paste a batch of text data for review here]
‘Hot signs’ you need to look for include:
1. Contradictory speech signals (internal statements that contradict public statements or known facts).
2. Negative emotional signals (words containing emotions such as panic, anger, and rush to cover up).
3. Abnormal command signals (e.g., ‘Delete immediately’, ‘Do not talk in company email’).
4. Acknowledgment or awareness of the signal (explicit or implicit acknowledgment of a problem or violation).
By doing so, you are putting on a pair of “special glasses” for the AI, which greatly improves the efficiency of your detection of “smoking guns” from massive evidence.
05 Don’t let it translate the law, let it be a “translator” to speak human language
A common problem for our lawyers is that when we talk to our clients, we are full of French language.
Explaining legal issues is not about reading the law, but about translating it into a “vernacular” that the customer can understand and is related to his interests.
Next time you want to explain a complex problem to the customer, let AI help you “translate”:
You now play two roles: A is a rigorous lawyer and B is a metaphorical communicator.
First, as Character A, explain in technical language: [Enter the complex legal issue you want to explain here].
Then, immediately switch to Role B and ‘translate’ Role A’s explanation into a verbal communication script for [describe your client, e.g., an anxious minority shareholder who doesn’t know anything about the law]. Translation Requirements:
1. The use of legal terms is prohibited or must be explained with extremely simple metaphors.
2. Focus on what customers care about most: ‘What is the best/worst outcome?’ What should I do? How long and how much does it cost? ’
3. The tone should be calming, and the conclusion should be pragmatic, giving hope and managing expectations.
The essence of this prompt lies in “role-playing” and “target audience portrait.”
It forces AI to restructure information from the customer’s point of view, which is a hundred times more effective than your dry interpretation of the law.
The prompts shown in this article are primary prompts, which are generally enough for lawyers to use! It is necessary to accurately operate related case projects according to different case situations and their own style, but it is necessary to customize more complex and accurate prompt words to solve them.
