By Victoria Carrington
Co-Chair, Innovation in Law Practice Committee

Generative AI tools can be useful in legal practice, but they are only as effective as the instructions lawyers provide. A vague prompt often produces a vague, incomplete, or unreliable response. A clear prompt can produce a more useful starting point.

Prompt engineering is not coding. For lawyers, it simply means giving an AI tool clear, structured, and appropriately limited instructions.

A helpful framework is RULES:

R: Role and Jurisdiction

Tell the AI what role it should assume and what jurisdiction applies.

For example:

You are a Utah employment attorney summarizing the enforceability of employee non-compete agreements under Utah law.

AI tools do not automatically know whether Utah law, federal law, trial practice, appellate practice, or a particular circuit applies. If jurisdiction matters, include it.

U: Understand the Task

Be specific about what you want the tool to do. Words like “analyze” or “help” may be too broad. Stronger prompts use concrete verbs such as:

• Draft
• Summarize
• Compare
• Rewrite
• Identify
• Explain

For example:

Summarize the following contract provision in plain English for a non-lawyer client.

L: Limits

Set boundaries. Without limits, AI tools may generate overly long, unfocused, or unusable responses.

Useful limits may include:

• Word count
• Tone
• Audience
• Format
• Number of examples
• Citation requirements

You can also instruct the tool to flag uncertainty rather than guess. This does not eliminate the risk of hallucinations, but it can help reduce unsupported or overconfident responses.

E: Expected Format

Tell the AI how you want the answer presented.

Examples include:

Provide the answer as a checklist.

Draft a client email in fewer than 150 words.

Create a table comparing the two provisions.

The more clearly you define the desired output, the easier it is to review and revise.

S: Say When It Is Wrong

AI output should be reviewed critically. If the tool misunderstands the task, applies the wrong assumption, or produces an inaccurate statement, correct it.

Lawyers remain responsible for the final work product. Treat AI-generated material like junior work product: potentially useful but requiring careful supervision.

Ethical Considerations

Generative AI does not replace a lawyer’s professional judgment. Lawyers using AI should continue to:

• Verify citations and legal authorities.
• Confirm legal conclusions.
• Protect client confidentiality.
• Follow applicable firm policies and client requirements.
• Review all AI-generated work before relying on it.

Do not input confidential client information into an AI tool unless you understand the tool’s privacy protections and your use complies with applicable professional obligations.

Final Takeaway

The lawyers who benefit most from generative AI will not necessarily be those who use it most often. They will be those who instruct it most effectively.

Better prompts lead to better drafts, more focused research assistance, and more efficient review. When using AI, remember the RULES: define the role and jurisdiction, explain the task, set limits, specify the expected format, and correct the tool when it is wrong.

*AI produced this blog from the transcript of the video.