Examples of prompting in legal practice
In legal practice, the ability to craft effective prompts for generative AI is becoming an essential skill. Legal prompting refers to the strategic construction of queries to AI systems in order to obtain accurate, relevant, and jurisdiction-specific legal content.
When done well, prompting can significantly improve the quality of outputs for tasks such as legal research, drafting, client communication, and risk analysis.
This document sets out some examples of prompting in legal practice to support varied legal use cases.
Key principles to effective legal prompting
1. Use clear and specific language
Avoid ambiguity. Be explicit about what you want, and phrase the query as if speaking to a junior colleague unfamiliar with the matter.
- Vague: “What is the law?”
- Specific: “Explain the core principles of contract formation under English law, including offer, acceptance, and consideration”.
2. Narrow the scope
Broad or open-ended questions often result in superficial responses. Narrowing the subject ensures relevance and improves the quality of the AI’s response.
- Too broad: “Tell me about employment law”.
- Improved: “Summarise an employer’s legal obligations when making redundancies under the Employment Rights Act 1996”.
3. Provide legal and factual context
AI performs better when it understands the circumstances of the request. Providing context – such as the purpose, user, or relevant legal developments – allows the AI to tailor its output more appropriately.
- Example: “Given the 2023 ICO guidance on marketing emails, explain how a UK charity can ensure compliance with data protection rules when sending newsletters”.
4. Define the desired output format
Specify how you want the information presented, for example, in bullet points, a plain-English summary, a clause-by-clause draft, or a comparison table. This avoids reformatting and aligns the output with its intended use.
- Example: “Provide a bullet-point summary of the key legal duties of charity trustees under the Charities Act 2011”.
5. Identify the audience or purpose
Stating the intended audience helps shape the tone and detail level. Whether for internal legal review, client education, or compliance documentation, the same legal issue can be addressed differently depending on the user.
- Example: “You are a solicitor advising a first-time homebuyer. Explain the key steps in the conveyancing process in plain English”.
| Use case | Objective | High-quality AI prompt | Notes / output format |
|---|---|---|---|
| Legal research | Retrieve recent case law on a specific topic | “List and summarise five recent UK cases (2018–present) that have clarified directors' duties under the Companies Act 2006. Include the case name, court, year, and one key legal takeaway from each.” | Bullet points or table format; citations included |
| Contract drafting |
Generate a standard commercial contract |
“Draft a commercial agreement for the sale of goods under English law. Include clauses for payment terms, delivery obligations, risk transfer, warranties, and governing law.” |
Clause-by-clause layout; use formal legal tone |
| Client communication | Translate legal content into plain English for non-lawyers | “You are a solicitor writing to a small business owner. Explain their obligations under the UK GDPR regarding employee data retention, in no more than 200 words, using plain English.” | Short paragraph; avoid legal jargon; aim for clarity |
| Litigation strategy | Generate risk insights based on past cases | “Identify three common legal risks in breach of contract claims under English law, supported by relevant case law examples. Summarise each risk and the circumstances in which it typically arises.” | Numbered list or bullet points; each point includes legal authority |
| Policy comparison | Compare legal regimes across jurisdictions | “Compare UK and EU consumer protection laws with a focus on refund and cancellation rights. Present differences in a side-by-side table, highlighting statutory sources and enforcement bodies.” | Table format; jurisdictional headings; include statute references |
| Legal training | Educate junior lawyers or students on core legal concepts | “Explain the distinction between legal and equitable remedies in English contract law. Provide definitions, key case examples, and one practical scenario where each would apply.” | Paragraph summary with embedded references to FCA source materials |
| Regulatory advice | Provide summarised compliance guidance | “Summarise the main FCA obligations for UK financial services firms regarding anti-money laundering (AML) procedures. Include references to the relevant Handbook rules and explain in clear, practical terms.” | Paragraph summary with embedded references to FCA source materials |
| Ethical guidance | Advise on emerging legal-ethical issues | “Outline three ethical risks of using generative AI tools in a legal practice setting. Explain each risk, its legal implications, and one recommended mitigation strategy.” | Numbered or bullet format; include clear explanation and practical steps |
| Document analysis | Analyse and improve an existing legal document or clause | “Review the following indemnity clause and identify any ambiguous or high-risk terms under UK law. Suggest revisions for clarity and legal effectiveness.” [Insert clause] | Two-column format: “Original wording” vs. “Suggested revision”, with rationale |
| Scenario drafting | Create realistic hypotheticals for case studies or internal training | “Generate a hypothetical legal dispute involving a breach of fiduciary duty by a company director under English law. Include background facts, the core legal issue, and possible remedies a claimant might seek.” | Multi-paragraph scenario with clear legal framing |
| Prompt rewriting | Improve a vague or ineffective user prompt | “Please rewrite the following prompt to be more specific, contextual, and targeted for legal research purposes. Original prompt: ‘Tell me about employment law’.” | Side-by-side display of “Original” and “Rewritten”; highlight changes and rationale |
| Client FAQ automation | Generate chatbot-ready answers to common client queries | “You are creating automated responses for a legal chatbot. Provide a concise, accurate answer (max 100 words) to the question: ‘What should I do if my landlord refuses to return my deposit?’ Ensure the response is aligned with current UK tenancy deposit regulations.” | Q&A format; limit word count; suitable for integration into helpdesk or website tools |
|
| Go back to the main course content |
|---|
