Prompt Triage

Purpose

This exercise helps students treat prompting as legal work planning, not as a search-box trick.

Students begin with a weak or underspecified prompt and diagnose why it is likely to produce generic, unreliable, or misaligned output. They then revise the prompt by identifying the task type, tool choice, missing context, constraints, output format, and verification plan.

The point is not to teach students a single “perfect prompt.” The point is to help them build a repeatable triage habit: before asking AI for help, pause long enough to decide what kind of help is actually needed and what professional guardrails should shape the output.

Learning Goals

By the end of this exercise, students should be able to:

1. Distinguish among research, analysis, drafting, editing, and validation prompts.
2. Identify when a general-purpose AI tool, legal-specific AI tool, traditional research database, or editing tool is the better fit for the task.
3. Diagnose what is missing from a weak prompt, including jurisdiction, audience, procedural posture, timeframe, facts, authority scope, and output format.
4. Add constraints that reduce risk, including facts-only instructions, no-speculation language, assumption lists, missing-information flags, tone requirements, and verification steps.
5. Revise a weak prompt into a stronger staged prompt that produces a more usable first output.
6. Explain what a lawyer would still need to verify before relying on the output.

Materials

• Three to six weak legal prompts for students to triage.
• A prompt triage worksheet with columns for task type, tool choice, missing context, needed constraints, output format, self-critique instruction, and verification step.
• A reusable prompt framework with fields for role, audience, task, inputs or facts, constraints, output format, self-critique, and open questions.
• Optional: access to AI tools so students can test the weak and revised prompts and compare outputs.

Setup

This exercise works best after students have been introduced to the idea that prompting is a form of supervision.

Students should understand that different legal tasks require different instructions. A research-orientation prompt is different from a drafting prompt. A validation prompt is different from a synthesis prompt. A general-purpose AI tool may require more scaffolding than a legal-specific tool, and a traditional research database may be the better choice when the task requires verified authority rather than generated explanation.

The instructor should frame the exercise around the lawyer’s supervisory role. A prompt is not just a request for text. It is an assignment: who is the AI acting as, who is the audience, what is the deliverable, what facts may be used, what must not be assumed, and what should be checked before anyone relies on the result?

Run of Show

Total, approximately 30–45 minutes

1. Frame the task Explain that students are not trying to write clever prompts. They are learning to diagnose weak prompts before using them, deciding what the prompt is actually asking for and what legal, factual, and professional constraints need to be added.
2. Prompt triage Give students several weak prompts. For each one, students identify the task type: research, analysis, drafting, editing, validation, or some combination.
3. Tool match Students decide whether the task is best suited for a general-purpose AI tool, a legal-specific AI tool, a traditional legal research platform, an editing tool, both AI and traditional research, or neither.
4. Missing-context diagnosis Students mark what information is missing, jurisdiction, court level, timeframe, procedural posture, client posture, audience, document type, relevant facts, source scope, and desired format.
5. Constraint building Students add guardrails: facts-only instructions, no speculation, neutral tone, length limits, required headings, assumption lists, missing-information flags, and directions to identify what must be verified.
6. Prompt revision Students revise one weak prompt into a stronger prompt that includes role, audience, task, inputs, constraints, output format, self-critique, and open questions.
7. Verification plan Students identify what they would verify before relying on the output, checking citations, confirming holdings, locating current statutes, reviewing procedural rules, validating facts against the record, or asking a follow-up prompt designed to surface weaknesses.
8. Debrief Discuss how the revised prompts differ from the originals. The output is not merely more detailed; it is more supervised, more constrained, and easier to evaluate.

Student Instructions

Instructor Notes

The central teaching point is that a weak prompt often reflects an unmade legal judgment.

Students may initially try to “make the prompt better” by adding more words. Push them to be more precise. The issue is not length. The issue is whether the prompt tells the tool what legal role it is playing, what task it is performing, what facts it may use, what assumptions it must avoid, what format is needed, and what the lawyer will verify afterward.

A weak prompt often fails because it asks AI to do too many things at once. “Write a memo about the contract dispute” asks the tool to identify the legal issue, infer the audience, choose a structure, invent missing facts, determine the jurisdiction, and draft final prose all at once. A stronger workflow stages the work: first generate a roadmap, then identify required elements, then draft one section, then critique, then revise.

Common weak-prompt problems include:

• No jurisdiction.
• No audience.
• No procedural posture.
• No client objective.
• No source scope.
• No facts-only instruction.
• No distinction between research and drafting.
• No instruction to flag assumptions.
• No verification plan.
• No indication of what “good” output looks like.

The strongest student revisions usually do three things: they narrow the task, they add professional constraints, and they make the next verification step visible.

Sample weak prompt: “Write a memo about the contract dispute.”

Sample stronger prompt: “Act as a junior associate preparing an internal research memo for a supervising partner. Using only the facts provided below, draft a roadmap for analyzing whether the seller breached the termination provision of the contract under Tennessee law. Do not add facts or speculate. Use headings for elements, likely arguments, factual gaps, and authorities to verify. At the end, list any assumptions you made and identify the next three research steps before drafting the full memo.”

The revised prompt is stronger not because it is longer, but because it assigns a role, identifies an audience, narrows the task, sets jurisdictional and factual limits, requires structure, and builds in verification.

Adaptation

Here are some suggestions on how to adapt this in other teaching contexts. The adaptations below have been limited to firm trainings and asynchronous suggestions, but there are other possible adaptations for workshops, seminars, and other contexts. Be creative!