Lesson 1.3: The Legal Research Prompt Pattern

Welcome to the cornerstone of your AI-powered legal research journey. In this lesson, we will dissect the single most important skill for leveraging Microsoft Copilot effectively: prompt engineering. The quality of your research output is directly proportional to the quality of your input. A vague or poorly constructed prompt will yield generic, often unhelpful, results. In contrast, a precise, well-structured prompt can unlock a level of detail and efficiency that transforms your practice. This lesson introduces a powerful, repeatable framework—the Jurisdiction + Scope + Task + Format (JSTF) pattern—that will serve as your foundation for all future legal research with Copilot.

Mastering this pattern is not merely a technical exercise; it is about shifting your mindset from asking simple questions to giving precise instructions. As an attorney or paralegal, you are already an expert in structuring logical arguments and requests. Prompt engineering applies that same rigor to your interaction with AI. By the end of this lesson, you will be able to deconstruct complex legal research questions into their core components, craft powerful prompts that deliver actionable results, and refine your inquiries for maximum accuracy and relevance. This is the skill that separates a novice user from a true power user, enabling you to harness Copilot as a reliable, efficient, and indispensable legal research assistant.

Deconstructing the Prompt Pattern: Jurisdiction + Scope + Task + Format

The JSTF pattern is a simple yet profoundly effective method for structuring your legal research prompts. It forces you to define the critical parameters of your request, leaving no room for ambiguity. Think of it as creating a detailed work order for your AI assistant. Each component builds upon the last, creating a logical sequence that guides Copilot to the precise information you need. Let’s explore each element in detail.

1. Jurisdiction: The Foundation of Your Query

The first and most critical element of any legal prompt is Jurisdiction. Law is geographically and constitutionally defined. A correct answer in one state is an incorrect answer in another. Failing to specify the jurisdiction is the most common mistake novice users make, and it is the fastest way to get unreliable or irrelevant results. Your prompt must explicitly state the governing law, whether it’s federal, a specific state, or a particular regulatory framework.

Why is this so important? AI models like Copilot are trained on a vast corpus of text, including legal documents from countless jurisdictions. Without a clear directive, the model may synthesize information from multiple sources, potentially blending conflicting legal standards. By starting with the jurisdiction, you immediately narrow the universe of relevant information, focusing the AI on the precise body of law that applies to your matter.

• State Law: When your issue is governed by state law, be explicit. Use phrases like “Under California law…”, “In Texas…”, or “Applying Florida statutes…”
• Federal Law: For issues of federal law, specify the context. Examples include “Under the Federal Rules of Civil Procedure…”, “Applying the Lanham Act…”, or “In the Ninth Circuit Court of Appeals…”
• Specific Regulatory Bodies: For administrative law questions, name the agency. For instance, “According to OSHA regulations…” or “Under the ABA Model Rules of Professional Conduct…”

Example: Instead of asking, “What are the elements of a breach of contract?” a much stronger prompt begins, “Under New York law, what are the elements of a breach of contract?”

2. Scope: Defining the Legal Issue

Once you have established the jurisdiction, the next step is to define the Scope of your legal inquiry. This component narrows the focus from a broad area of law to the specific legal question or issue you are researching. A well-defined scope prevents the AI from providing overly broad or tangential information. It is the difference between asking for a map of the entire country and asking for directions to a specific address.

Think of the scope as the “question presented” in a legal brief. It should be a concise statement of the precise legal doctrine, claim, or issue at hand. The more specific you are, the more targeted and useful the response will be. This is where your legal expertise comes into play, as you must accurately frame the issue to get a relevant answer.

• Specify the Cause of Action: Instead of a general area like “torts,” specify “for a claim of negligence…” or “regarding the tort of intentional infliction of emotional distress…”
• Reference a Specific Standard or Rule: If you are researching a procedural question, name the rule. For example, “…concerning the standard for a preliminary injunction…” or “…regarding the admissibility of hearsay evidence under Rule 803…”
• Focus on a Specific Element: For complex claims, you can narrow the scope to a single element. For instance, “…focusing on the element of causation for a medical malpractice claim…”

Example: Building on our previous prompt, we add scope: “Under New York law, for a breach of contract claim involving the sale of goods, what are the elements…?” This is far more precise than simply asking about breach of contract generally.

3. Task: Stating Your Exact Need

With jurisdiction and scope established, you must now define the Task. This is the verb of your prompt—the explicit instruction telling Copilot exactly what you want it to do with the information. Are you asking for a summary, a comparison, a list, or a draft? Without a clear task, the AI is left to guess, and it may provide a narrative summary when what you really needed was a checklist.

Using strong, action-oriented verbs is crucial here. Avoid passive or ambiguous language. The task should be an unambiguous command. This is how you control the nature and structure of the AI’s output, ensuring it aligns with your specific research objective.

• Summarize: Use this for condensing information. “Summarize the key holdings…” or “Summarize the arguments for and against…”
• Identify/List: This is ideal for extracting specific pieces of information. “Identify the statutory factors…” or “List the elements a plaintiff must prove…”
• Compare/Contrast: Use this for analytical tasks. “Compare the requirements for a valid will versus a living trust…” or “Contrast the discovery rules in federal court with those in [State] court…”
• Draft: This is for generating initial work product. “Draft a set of interrogatories…” or “Draft a client email explaining…”
• Explain: Use this for conceptual understanding. “Explain the doctrine of res judicata…”

Example: Continuing our prompt: “Under New York law, for a breach of contract claim involving the sale of goods, identify the essential elements a plaintiff must plead and prove…” This task is clear and direct, telling Copilot to extract and list specific information.

4. Format: Structuring the Output

The final component of the JSTF pattern is Format. This instruction tells Copilot how to structure and present the information it returns. It is the key to transforming a dense block of text into a well-organized, easy-to-scan, and immediately useful work product. Failing to specify the format often results in a narrative-style answer that requires significant time to parse and re-organize. By defining the format, you are essentially creating a template for the AI to fill in.

A well-chosen format can make the difference between a research session that feels like a conversation and one that feels like a productive work session. It helps you quickly assess the information and integrate it into your workflow, whether you are drafting a memo, preparing for a client meeting, or creating a case strategy outline.

• Numbered List: This is perfect for sequential steps, elements of a claim, or statutory factors. “…in a numbered list.”
• Bulleted List: Use for non-sequential items, such as a list of potential arguments or relevant documents. “…in a bulleted list.”
• Comparison Table: This is exceptionally powerful for analyzing two or more concepts side-by-side. “…in a comparison table with columns for [Concept A], [Concept B], and [Key Differences].”
• Formal Memo Format: When you need to generate a first draft of a document, specify the structure. “…in a formal memo format with sections for Question Presented, Brief Answer, Discussion, and Conclusion.”
• Email Draft: For client communications, this provides a ready-to-edit starting point. “…as a draft email to a client.”

Example: Completing our prompt: “Under New York law, for a breach of contract claim involving the sale of goods, identify the essential elements a plaintiff must plead and prove, and present them in a numbered list with supporting case citations for each element.” This final instruction ensures the output is not just correct, but also perfectly structured for use in a research memo or pleading.

Putting It All Together: Good vs. Bad Prompts

Understanding the theory behind the JSTF pattern is one thing; seeing it in action is another. The difference between a well-crafted prompt and a poor one is stark. A bad prompt is typically short, ambiguous, and lacks the specific components of the JSTF pattern. It forces the AI to make assumptions, which often leads to generic, irrelevant, or even incorrect information. A good prompt, by contrast, is a detailed instruction that leaves nothing to chance. It is specific, clear, and follows the JSTF pattern to guide the AI to the precise answer you need.

Let’s examine some side-by-side comparisons to make this distinction crystal clear. Notice how the “Good Prompt” examples explicitly incorporate Jurisdiction, Scope, Task, and Format, while the “Bad Prompt” examples are vague and open-ended.

Example Prompts by Practice Area

The JSTF pattern is a universal framework that can be adapted to any area of legal practice. The key is to customize the Jurisdiction and Scope to fit the specific context of your matter. Below is a table of example prompts across various practice areas. Use these as a starting point for your own research tasks. Notice how each one, while different in subject matter, adheres strictly to the JSTF pattern to ensure clarity and precision.

Beyond the First Prompt: Refinement and Iterative Prompting

A perfect prompt is not always achievable on the first try. Legal research is often a process of discovery, where the answer to one question leads to another. Effective use of Copilot embraces this reality through iterative prompting. This is the practice of using the AI’s initial response to refine your follow-up questions, creating a conversational workflow that allows you to drill down into a topic with increasing specificity. Think of it as conducting a witness examination; you start with broad questions and use the answers to ask more targeted, probing follow-up questions.

The iterative process typically involves adjusting one or more components of the JSTF pattern based on the information you receive. You might narrow the Scope, change the Task, or request a different Format to explore the issue from a new angle.

The Iterative Prompting Workflow

1. Start with a Broad JSTF Prompt: Begin with a well-structured but relatively broad prompt to establish a foundational understanding of the legal landscape.
2. Analyze the Initial Response: Carefully review the AI’s output. Does it answer your question directly? Does it introduce new concepts, cases, or statutes that you were not aware of? Does it reveal ambiguities in your initial query?
3. Formulate a Follow-Up Prompt: Craft a new prompt that builds directly on the previous response. This can be done in several ways:
   • Narrowing the Scope: Focus on a specific aspect of the initial response. If the first prompt was about the elements of negligence, a follow-up might focus solely on the element of causation.
   • Changing the Task: Ask the AI to perform a different operation on the same information. If you first asked for a summary, you might next ask for a comparison or a draft.
   • Requesting Clarification: If the AI’s response is unclear or uses unfamiliar terminology, ask for an explanation.
   • Expanding the Jurisdiction: You might ask how the law in a neighboring state compares to the information you have already received.
4. Repeat as Necessary: Continue this cycle of prompting and refining until you have the precise level of detail and understanding you need.

Example of an Iterative Prompting Session

Let’s walk through a scenario to see how this works in practice. Imagine you are researching the requirements for personal jurisdiction over an out-of-state defendant.

Initial Prompt (Broad JSTF):

Under federal law, concerning a federal court’s personal jurisdiction over an out-of-state corporate defendant, summarize the constitutional “minimum contacts” test established in International Shoe and its progeny, in a paragraph.

Copilot’s Potential Response: The AI provides a good summary of the general and specific jurisdiction standards, explaining that general jurisdiction requires the defendant to be “at home” in the forum state, while specific jurisdiction arises from the defendant’s specific contacts with the state that gave rise to the claim.

Follow-Up Prompt 1 (Narrowing the Scope): The response mentioned “specific jurisdiction,” which is most relevant to your case. You now drill down on that concept.

Regarding the “specific jurisdiction” part of your answer, list the three elements of the Ninth Circuit’s test for analyzing specific jurisdiction. Present this as a numbered list.

Copilot’s Potential Response: The AI lists the three-prong test: (1) The defendant must have purposefully directed activities or consummated some transaction with the forum or a resident thereof; (2) the claim must arise out of or relate to the defendant’s forum-related activities; and (3) the exercise of jurisdiction must be reasonable.

Follow-Up Prompt 2 (Changing the Task & Format): The response mentions the “purposeful direction” element, which is the key issue in your fact pattern. You now want to see how this applies in a real-world context.

Using the first element of that test, “purposeful direction,” provide three examples of actions a corporate defendant might take that would satisfy this prong in a trademark infringement case. Format this as a bulleted list.

Copilot’s Potential Response: The AI provides examples, such as operating a commercial website accessible to residents of the forum state, shipping infringing goods directly to customers in the state, or running a targeted advertising campaign aimed at the state’s residents.

Through this three-step iterative process, you have moved from a general doctrinal question to highly specific, fact-pattern-relevant examples. This is far more effective than trying to craft a single, perfect, and overly complex prompt at the outset. It allows your research strategy to evolve as you learn.

Conclusion: From Prompting to Practice

The Legal Research Prompt Pattern—Jurisdiction, Scope, Task, and Format—is more than just a formula; it is a new way of thinking about legal research in the age of AI. By internalizing this pattern, you transform Microsoft Copilot from a simple search engine into a sophisticated research assistant capable of understanding and executing complex instructions. This structured approach ensures that you remain in control of the research process, guiding the AI to deliver the precise, relevant, and well-organized information your practice demands.

As you move forward, remember that prompt engineering is a skill that improves with practice. Continuously challenge yourself to be more specific in your scope, more deliberate in your task instructions, and more creative in your formatting requests. Embrace the power of iterative prompting to peel back the layers of complex legal issues, allowing your inquiry to evolve with your understanding. By mastering the techniques covered in this lesson, you are not just learning to use a new tool; you are developing a core competency that will define the modern, efficient, and competitive legal professional for years to come.