A few practical techniques can help legal professionals get more relevant, focused, and useful results from legal AI.
As legal AI becomes more familiar, the conversation is moving beyond whether lawyers should use these tools to how they can use them more effectively.
Prompts don’t have to be complex, and we don’t need intensive training; however, structuring the prompt effectively will make all the difference,
In many cases, improving the quality of an AI-generated response comes down to providing clearer context, defining the task more precisely, and telling the system what you expect from the output.
For legal professionals, this is particularly important.
A well-structured prompt can help focus the AI on the relevant jurisdiction, legal issue, document or perspective, making the resulting output easier to evaluate and use.
Prompting should be adapted, depending on the task and the desired output. Here are some guidelines to improve your output;
1. Add a POV to your prompt
One of the simplest ways to add context to a prompt is to tell the AI which perspective it should adopt.
Known as a persona prompt, this technique can help produce an answer that is more focused on the relevant legal context, audience or issue.
For example:
Prompt:
Act as a tribunal in a DIAC arbitration seated in Dubai hearing a complex commercial dispute. Provide a concise and focused summary of the tribunal’s approach to immediate judgment applications.
Or:
Prompt:
You are senior UAE insurance defence counsel with deep expertise in corporate governance and shareholder disputes. Explain how to structure a strong claim on behalf of a minority shareholder of a UAE-incorporated company.
The more precisely the role is defined, the more context the AI has when generating its response.
However, a persona doesn’t make the AI an expert. The legal professional must still assess whether the answer is accurate, appropriate and supported by authoritative sources.
2. Use documents to focus the analysis
Legal work often depends on identifying relationships between facts, dates, clauses, correspondence and other documents.
Where a legal AI tool allows documents to be uploaded and analysed, prompts can be used to direct attention towards a particular issue or task.
For example:
Prompt:
Review the attached documents and provide a detailed timeline of events relevant to Party X’s employment. Present the response in a table that identifies the fact, date, document source, and relevance.
Or:
Prompt:
Review the uploaded contract and provide an objective risk assessment focusing on Party Y’s interests and the commercial and legal risks that may require further review.
Another useful approach is to ask the AI to consider the documents from an opposing perspective:
Prompt:
Act as respondent’s counsel and identify potential weaknesses in the uploaded claim. Explain which issues may require further factual or legal investigation.
These prompts can help legal professionals interrogate large volumes of information more efficiently. They shouldn’t replace a lawyer’s own review of the underlying documents, particularly where accuracy, confidentiality or legal strategy is involved.
3. Ask for the authority behind the answer
A legal AI response becomes significantly more useful when the legal professional can identify and verify the sources supporting it.
Where the tool has access to reliable legal sources, the prompt should make that requirement explicit.
For example:
Prompt:
Review this paragraph and assess whether the arguments and authorities are supported under Saudi law. Identify the applicable legislation, regulations and relevant court or arbitral decisions where available.
Or:
Prompt:
Provide a summary of commentary concerning the A v B decision and identify the source supporting each material point.
This approach is particularly useful during legal research because it helps move the task beyond simply generating an answer.
It creates a starting point for verification.
The presence of a citation, however, shouldn’t be treated as proof that the answer is correct. Lawyers should confirm that the authority exists, that it says what the AI claims it says, that it remains current and that it is appropriate for the relevant jurisdiction and matter.
4. Use AI as an editor
Legal drafting often involves several rounds of refinement.
The substance of an argument may already be sound, while the structure, tone or length still needs work. Legal AI can be useful at this stage because the lawyer can define exactly what should change and, equally importantly, what should remain untouched.
For example:
Prompt:
Without changing the legal analysis, rewrite this argument to be approximately 30% shorter and more direct. Preserve all authorities and citations.
Or:
Prompt:
Edit this email to opposing counsel so that it is firm, professional and non-inflammatory without softening the legal position.
For longer submissions:
Prompt:
Reorganise this section using clear headings and subheadings so that the key issues and arguments can be identified quickly by a busy judge or arbitrator. Do not change the underlying legal analysis.
Clear instructions help preserve the lawyer’s judgement and reasoning while allowing the AI to assist with presentation and structure.
5. Turn complex workflows into practical checklists
Legal work is often process-driven. Important steps can be missed when matters become complex, deadlines overlap or several people are involved.
A legal AI tool can help create an initial checklist or workflow for the lawyer to review and adapt.
For example:
Prompt:
Create a step-by-step checklist for defending an unjust dismissal claim in the financial services sector, from initial intake through to settlement. Identify points at which jurisdiction-specific legal advice or procedural verification may be required.
Or:
Prompt:
Prepare an outline and proposed questions for interviews with the complainant, respondent and witnesses in a workplace harassment investigation. Where appropriate, explain the purpose of each category of questions.
For document-heavy matters:
Prompt:
Prepare a draft document production plan for a commercial lease dispute. Organise it by legal issue and include relevant document categories and potential lines of inquiry for each party.
The output should be treated as a working framework rather than a definitive legal process. Lawyers should check it against the relevant law, procedural requirements, matter strategy and professional obligations.
Better prompting starts with better instructions
- Providing a role can sharpen perspective.
- Supplying documents can focus the analysis.
- Asking for authority makes verification easier.
- Editing prompts can improve communication.
- Checklist prompts can bring more structure to complex workflows
These techniques can also be combined.
A legal professional might ask the AI to review uploaded documents from the perspective of opposing counsel, identify potential weaknesses, organise them by issue, and provide the legal authorities to be checked.
The possibilities are broad, but the professional responsibilities remain the same.
Legal AI can support research, analysis, drafting and workflow. The lawyer still needs to review the output, verify the underlying sources and apply professional judgment before relying on it.
The most effective prompt is therefore not necessarily the longest or most sophisticated.
It’s the one that gives the AI enough context to be useful while keeping the legal professional firmly in control.