Hands-On Exercises: Putting Theory into Practice

Theory and concepts are essential, but true mastery of a new skill comes from hands-on application. This lesson is dedicated to moving beyond discussion and into direct, practical engagement with Microsoft Copilot in a simulated legal environment. You will now have the opportunity to apply the principles of scope control and the Scope + Task + Format prompting pattern to perform the core tasks of summarization, extraction, and comparison. We have prepared a mock matter environment within SharePoint, populated with a variety of typical legal documents, including pleadings, contracts, and correspondence. By working through these structured exercises, you will build muscle memory, gain confidence in your ability to command the AI, and experience firsthand the transformative impact this tool can have on your daily workflow.

The goal of this session is not to test you, but to empower you. It is a safe space to experiment, make mistakes, and learn by doing. As you work through the prompts, pay close attention to how small changes in your wording can affect the output. Observe how narrowing your scope improves the relevance of the results. The objective is to internalize the structured prompting framework so that it becomes second nature. This practical experience is the bridge between understanding what Copilot can do and knowing how to make it do what you need it to do, consistently and reliably. Remember, every prompt you write is an opportunity to refine your skill and become a more efficient and effective legal professional in the age of AI.

Preparing Your Workspace: The Mock Matter Environment

Before we begin, please navigate to the SharePoint site and specific document library provided by your instructor. This library, titled “Mock Matter – Project Apollo,” contains a curated set of documents that we will use for the following exercises. You will find a variety of file types, including PDF documents representing court filings and Word documents representing draft agreements. Take a moment to familiarize yourself with the file names and the overall structure of the library. Notice that the files have been named using a consistent convention (e.g., Apollo_v_Orion_Complaint_20251015.pdf) and that the library includes metadata columns for Document Type and Status, reflecting the Information Architecture principles we have discussed.

It is important to understand how these metadata columns enhance your interaction with documents and Copilot alike. For example, filtering the library by Document Type can immediately isolate pleadings, contracts, or correspondence. This allows you to set a precise scope for AI queries, which is crucial for obtaining relevant and actionable responses. As you become more comfortable, consider experimenting with additional filters or views that categorize documents by date, author, or jurisdiction, enabling even more targeted prompting.

Additionally, take note of the folder structure and naming conventions. Consistency here not only helps you navigate quickly but also supports Copilot’s ability to contextualize your queries. For instance, when you select multiple documents for summarization or comparison, Copilot uses file names and metadata context to infer relationships and content types, which can improve the quality of its output.

Practical Tip: Before starting any AI-driven task, spend a few minutes organizing your workspace. Confirm that the relevant documents are appropriately tagged and named. This upfront investment in Information Architecture will pay off in more precise AI results and less manual correction later.

Exercise 1: Summarizing Key Pleadings

In this exercise, you will use Copilot to quickly get up to speed on the core arguments of a new case by summarizing the initial complaint and the subsequent answer. This simulates the common task of being assigned to a new litigation matter and needing a rapid overview.

1. Select the Files: In the “Mock Matter – Project Apollo” library, find and select the following two files:
   • Apollo_v_Orion_Complaint_20251015.pdf
   • Apollo_v_Orion_Answer_20251105.pdf
2. Launch Copilot: With both files selected, click the Copilot button in the SharePoint command bar. This action sets your scope to only these two documents.
3. Craft Your Prompt: In the Copilot pane, type the following structured prompt:
   

   “Using the selected files, summarize the key allegations made in the Complaint and the corresponding defenses raised in the Answer. Present the summary in two distinct sections with clear headings for each document.”

4. Analyze the Output: Review the summary generated by Copilot. Does it accurately capture the main points of both the complaint and the answer? Does the two-section format make the information easy to digest? This exercise demonstrates how quickly you can gain a balanced understanding of the core legal dispute in a matter.

Expanding on this exercise, consider how you might customize the prompt to fit your specific needs. For example, you might ask Copilot to highlight any jurisdictional or procedural issues mentioned in the pleadings. Or you could request a timeline of key events extracted from the documents to assist in case strategy discussions.

Practical Tip: When summarizing pleadings, explicitly requesting separation of allegations and defenses helps maintain clarity. Legal documents often contain dense language and multiple claims; breaking the summary into sections ensures that critical information is not lost or conflated.

Example Prompt Variation:

“Summarize the Complaint and Answer, highlighting the factual allegations, legal claims, defenses, and any affirmative defenses. Include a brief overview of the procedural posture.”

This encourages Copilot to produce a more detailed and multi-layered summary, which may be beneficial in complex matters or when briefing senior attorneys.

Additional Practical Application: In real-world legal practice, quickly understanding pleadings is vital for new team members or when preparing for client meetings. Using Copilot to generate initial summaries allows you to focus your substantive review and prepare more insightful questions or analyses for the case team.

Exercise 2: Extracting Key Terms from a Contract

This exercise focuses on structured data extraction. You will task Copilot with pulling specific business terms from a draft agreement and organizing them into a structured format. This is a common task in contract review and due diligence.

1. Select the File: In the library, deselect any previous files and select the single document named MSA_Draft_Stark_Industries_20260228.docx.
2. Launch Copilot: With the file selected, open the Copilot pane.
3. Craft Your Prompt: This time, you will ask for specific data points to be formatted into a table. Type the following prompt:
   

   “From the selected draft agreement, extract the initial term of the agreement, the payment schedule, the governing law, and the amount of the liability cap. Present this information as a table with two columns: ‘Term’ and ‘Details’.”

4. Analyze the Output: Examine the table produced by Copilot. Is the information accurate? Is it presented in the format you requested? This exercise highlights how Copilot can be used to create a quick reference sheet of key commercial terms from any contract, saving significant time during due diligence or contract review.

To deepen your understanding, reflect on the value of rapid term extraction in your own legal practice. For instance, during M&A due diligence, lawyers often sift through hundreds of contracts to verify material terms. Automating this extraction with Copilot can reduce review time dramatically, enabling you to focus on risk assessment and negotiation strategy.

Advanced Tip: You can extend this approach by requesting additional contract elements such as renewal terms, termination rights, confidentiality clauses, or dispute resolution mechanisms. Structuring your prompt with a clear list of terms ensures Copilot knows exactly what to look for.

Example Prompt for Comprehensive Extraction:

“Extract the following key terms from the selected agreement: initial term, renewal options, payment schedule, termination rights, governing law, liability cap, confidentiality obligations, and dispute resolution provisions. Present the data in a table with ‘Term’ and ‘Details’ columns.”

Such a table becomes a powerful tool for quick reference, internal reporting, or client updates, improving communication and reducing the risk of missing critical contract elements.

Practical Tip: When you receive the table, verify the accuracy by spot-checking terms against the document. AI tools can occasionally misinterpret complex or ambiguously worded provisions, so professional oversight remains essential. Use this as a starting point rather than a final deliverable.

Additional Benefit: Copilot’s ability to output in structured formats like tables or bullet points is particularly useful for integrating AI insights into your legal project management tools or spreadsheets, enabling seamless data analysis and reporting.

Exercise 3: Comparing Two Contract Drafts

This exercise applies the document comparison skill. You will ask Copilot to identify the key differences between two versions of a contract, simulating the common task of reviewing redlines from opposing counsel.

1. Select the Files: In the library, deselect any previous files and select the following two documents:
   • MSA_Draft_Stark_Industries_20260228.docx
   • MSA_Draft_Stark_Industries_20260301_Redline.docx
2. Launch Copilot: With both files selected, open the Copilot pane.
3. Craft Your Prompt: This time, use a focused comparison prompt. Type the following:
   

   “Compare the two selected drafts of the Stark Industries MSA. Focus specifically on any changes to the sections related to the liability cap, the indemnification obligations, and the governing law. Summarize the differences in a table with columns for ‘Clause’, ‘Original Version’, and ‘Revised Version’.”

4. Analyze the Output: Review the comparison table. Does it accurately identify the key changes? Is the table format easy to read and useful for a negotiation discussion? This exercise demonstrates how a focused comparison prompt can give you an immediate, strategic understanding of the most important changes in a redlined document.

Understanding contract changes is crucial to mitigating legal risk and advising clients effectively. This exercise mirrors the real-world scenario where opposing counsel sends revised drafts, and you must quickly identify and evaluate substantive changes. By leveraging Copilot, you can accelerate this review process significantly.

Pro Tip: When dealing with complex contracts, you can break down your comparison prompts into smaller sections, focusing on key thematic areas like payment terms, liability, or confidentiality. This prevents overwhelming the AI with large volumes of text and enhances output accuracy.

Example of a Multi-Part Prompt:

“First, compare the payment terms sections in both drafts and summarize any changes. Then, perform a similar comparison for liability and indemnification provisions. Present each comparison in separate tables.”

This segmented approach can improve clarity and make the results easier to review, especially when working with colleagues or clients who may only be interested in particular sections.

Additional Insight: Copilot can also assist in generating suggested language or alternative clauses based on the comparison. After identifying changes, you might prompt Copilot to propose negotiation points or risk mitigation strategies related to altered provisions.

Example Follow-Up Prompt:

“Based on the identified changes to the liability cap and indemnification sections, suggest alternative language that balances client protection with contractual fairness.”

This capability transforms Copilot from a passive reviewer into an active drafting assistant, enhancing your negotiation leverage.

Exercise 4: Experimenting with Scope

This final exercise is designed to illustrate the critical importance of scope control. You will run the same prompt at two different scope levels and observe how the results differ dramatically.

1. Step 1: Broad Scope. Navigate to the top level of the SharePoint site (not inside any specific library). Open the Copilot pane and type the following prompt:
   

   “Summarize the key issues in the Apollo v. Orion case.”

   Observe the result. It may be accurate, but it might also pull in information from other documents on the site that are not directly relevant to the case, or it might be overly general.

2. Step 2: Narrow Scope. Now, navigate into the “Mock Matter – Project Apollo” library. Select only the two pleading files (Complaint and Answer). Open the Copilot pane and type the exact same prompt:
   

   “Summarize the key issues in the Apollo v. Orion case.”

   Observe the result. It should be significantly more focused, relevant, and accurate, because Copilot’s analysis is now grounded in only the two most relevant documents.

This exercise powerfully demonstrates the principle that scope is the most important lever you have for controlling the quality of Copilot’s output. The same question, asked at different scope levels, can produce vastly different results. By making it a habit to always narrow your scope before prompting, you will consistently achieve better, more reliable outcomes. This is the single most impactful takeaway from the entire hands-on session.

To further illustrate, consider how scope not only affects relevance but also influences the AI’s confidence in its responses. When the AI has a focused document set, it can cross-reference facts and language more effectively, reducing hallucinations or errors. When scope is too broad, Copilot may attempt to synthesize conflicting information or fill gaps with assumptions, which can lead to misleading or incomplete summaries.

Practical Advice: Always think critically about the documents you select before prompting. For example, if you are preparing a memo on contract risk, limit your scope to the relevant contract(s) and related correspondence, rather than including unrelated pleadings or discovery materials. This ensures that Copilot’s output is tightly aligned with your objective.

Moreover, consider the impact of scope on privacy and confidentiality. Narrowing your scope can help avoid inadvertently exposing sensitive information from unrelated documents when sharing AI-generated insights with clients or colleagues.

Finally, scope control is not static. As your inquiry evolves, you may need to broaden or narrow scope dynamically. For example, start with a narrow scope to get a detailed summary, then expand to include related documents for broader context or risk assessment.

Debrief and Discussion

Take a few minutes to reflect on your experience with these exercises. Consider the following questions:

• Which exercise did you find most immediately useful for your daily work?
• Were there any prompts where the output surprised you, either positively or negatively?
• How did the scope experiment in Exercise 4 change your understanding of how Copilot works?
• What other tasks in your practice could you now apply the Scope + Task + Format pattern to?

Share your observations with the group. Learning from each other’s experiences is one of the most valuable aspects of this workshop. The insights you gain from your colleagues’ experiments can help you discover new use cases and refine your own prompting techniques.

To expand the discussion, consider how Copilot’s capabilities might be integrated into broader legal workflows beyond document review. For example, in eDiscovery, Copilot could assist in identifying relevant documents by summarizing large data sets or extracting custodian information. In compliance monitoring, it could flag discrepancies between contractual obligations and actual performance data.

Moreover, think about how Copilot might support knowledge management within your firm. By summarizing legacy matters or extracting precedent clauses, AI can transform static repositories into dynamic, searchable knowledge bases that enhance firm-wide expertise.

Key Principle: Effective AI adoption in legal practice depends on combining technological tools with human judgment and domain expertise. Use Copilot to augment your work, not replace critical legal analysis.

Finally, consider the ethical and confidentiality implications of AI use in your practice. Always ensure that client data is handled in accordance with applicable rules and firm policies. Transparency with clients about AI use can build trust and demonstrate your commitment to leveraging technology responsibly.

As you continue to experiment with Copilot and SharePoint, keep in mind that mastery comes with practice, feedback, and continuous refinement of your prompting skills. Document your successful prompts and share best practices within your team to accelerate collective learning.