18 July 2026
10 ChatGPT prompts for lawyers (with examples)
ChatGPT does not draft pleadings for you, nor does it replace legal expertise, but used sensibly it saves hours on repetitive tasks: drafts, summaries, checklists, routine emails. The difference between a useful output and a worthless one lies in the prompt: the more specific it is about role, facts, format and tone, the closer the result is to something you can actually use in practice.
Below are 10 ready-made prompts to copy and adapt, organised by task: analysis, writing and client communication. A word of warning before you start: never enter real identifying details of clients (names, tax codes, case numbers, addresses) into prompts sent to public chatbots such as the consumer version of ChatGPT. Replace them with generic placeholders ("Mr Rossi" becomes "the claimant").
Prompts for analysing judgments, case files and contracts
1. Summarise a judgment
Prompt:
Act as a law firm assistant specialising in [civil/criminal/administrative] law. I am pasting in the text of a judgment (anonymised). Give me a structured summary in: 1) facts of the case, 2) grounds for the decision, 3) point of law established, 4) the order/ruling. Maximum 300 words, technical but concise language. Highlight whether the judgment departs from settled case law.
Paste the text of the ruling after the prompt, removing the names of the parties. Useful for internal case-law digests or for a quick note to the client.
2. Summarise a case file
Prompt:
I have a set of documents relating to a matter concerning [type of dispute]. Analyse them and produce a chronological summary of the relevant facts, a list of the missing documents that would be needed to complete the evidentiary picture, and the three most critical points for our procedural position. Flag any contradictions between the documents.
Good as a first screening before an internal meeting: it does not replace reading the file in full, but it helps you get your bearings on voluminous matters.
3. Look for counterarguments and rebuttals
Prompt:
I am arguing the following line of defence in a [practice area] case: [describe the argument in 3-4 lines]. List the main objections that an experienced opposing party might raise, order them by how dangerous they are, and for each one suggest a possible line of response. Do not invent legislative or case-law references: if you are not sure, say so explicitly.
A good "stress test" of the strategy before a hearing or a brief: it is there to show you where the argument is weak, not to copy the answers without checking them.
4. Review checklist for a contract
Prompt:
I need to review a draft contract of [type: commercial lease / supply / works contract / NDA]. Generate a review checklist with the essential clauses to verify (term, termination, penalties, jurisdiction/competent court, confidentiality, limitation of liability, etc.), highlighting which clauses are typically the most negotiated and in whose favour they tend to be skewed.
Use it as a reminder before the review proper, then paste in the individual clauses for a detailed comparison.
Prompts for drafting pleadings, opinions and emails
5. Draft of a formal notice to perform (letter before action)
Prompt:
Act as an Italian civil lawyer. Write a draft formal notice to perform (letter before action) for non-payment of a commercial invoice. Structure: heading, statement of facts (I will add the details), demand for payment within 15 days, reference to Articles 1453 and 1454 of the Italian Civil Code, reservation of all legal action and of a claim for default interest. Formal tone, short sentences, no real data: I will use placeholders such as [PARTY NAME] and [AMOUNT].
Let it generate the skeleton, then insert the specific facts and have the final text reread before sending.
6. Draft of a legal opinion (parere pro veritate)
Prompt:
I need to draft a legal opinion (parere pro veritate) on a [practice area] question. The issue is as follows: [describe the issue in anonymised form]. Structure the opinion into: the question, the applicable legal framework (indicate the sources only if you are reasonably certain of them, otherwise note that they need to be verified), analysis, operational conclusions with any risk rating (low/medium/high). Technical, expository tone.
Treat the output as a draft: every legislative or case-law reference must be checked against official sources before it goes into the final opinion.
7. Payment reminder email
Prompt:
Write a payment reminder email for a professional invoice overdue by 30 days. Firm but professional tone, not aggressive, referring to the possibility of applying default interest under Italian Legislative Decree 231/2002 in the event of further delay. Maximum length 120 words, suitable for a client with whom you want to maintain a professional relationship.
Ask for a second, more assertive version too, to use if the reminder goes unanswered.
8. Questions for a hearing or for examining a witness
Prompt:
I am preparing the examination of a witness in a [practice area] case. The witness will give evidence on [subject of the testimony, in brief]. Prepare a list of 10 questions, ordered logically from the general to the specific, avoiding leading questions, useful for bringing out the facts that support our argument.
Always review the order and wording according to the procedure: the model does not know the file, so the questions must be tailored to the real facts of the case.
Prompts for client communication
9. Translate legal language into plain language
Prompt:
I am pasting in an extract from a court document (or a contractual clause). Rewrite it in plain language, understandable to a client with no legal training, keeping the exact meaning and without omitting important practical consequences (deadlines, obligations, risks). Use short sentences and concrete examples where useful.
Perfect for accompanying the delivery of a document with an explanatory note, cutting down on clarification phone calls.
10. Summary of a client meeting
Prompt:
I am giving you the notes (even if disorganised) taken during a meeting with a client. Turn them into a concise minute with: date, attendees (I will use placeholders), points discussed, decisions taken, next steps with their deadlines and the person responsible for each step. Professional tone, bullet-point format.
Paste in the already-anonymised notes and in a few seconds you will have a draft minute to refine and send to the client.
If these examples convince you of the usefulness of structured prompting but you do not fancy rewriting a prompt from scratch every time, Tomstore's pay-per-use AI prompt generator includes ready-made templates for the legal field, already set up with the correct role, format and tone: you pay only for the prompts you generate, with no subscriptions.
Frequently asked questions
Can artificial intelligence replace a lawyer?
No. ChatGPT can speed up drafts, summaries and repetitive tasks, but it has no professional liability, does not know the client's real file and cannot sign a pleading or take on the professional duties of the engagement. It remains a support tool, not a substitute for the lawyer's judgment.
Can I trust the legislative or case-law citations generated by AI?
No, not without verification. Language models can generate plausible but non-existent articles of law or judgments, or cite the source correctly while getting its content wrong ("hallucinations"). Every reference must be checked against official databases before it goes into a pleading or an opinion.
Is it safe to paste a client's documents into ChatGPT?
With the free or consumer version, better not, or only after removing identifying details (names, tax codes, case numbers, addresses). It is a matter both of professional confidentiality and of personal data protection.
What are the most common mistakes when a lawyer starts using AI?
The most frequent ones: trusting the output blindly, writing prompts that are too generic and getting vague answers, and not specifying the desired format.
Is it worth always using the same prompt for similar tasks?
It is worth having a reusable base but adapting it each time to the specific facts: a prompt that is too rigid produces standardised answers that then require customisation anyway. Keeping a small library of tested prompts, with well-defined placeholders, is the most efficient way to avoid starting from scratch every time.