Lesson 3.2: Building and Querying Case File Agents

Welcome to Lesson 3.2 of Course 04: Copilot for the Modern U.S. Law Firm by legalGPTs. In this lesson, we will explore how to build and query Agents in OneDrive using Microsoft 365 Copilot—powerful AI assistants designed to transform the way law firms interact with their case files. By the end of this session, you will be equipped with the knowledge and practical skills to create dedicated AI agents grounded in specific case folders, enabling faster, smarter access to critical legal information.

1. What Are Agents in OneDrive?

Agents in OneDrive refer to dedicated AI assistants that are contextually grounded in particular files or folders within your OneDrive environment. These agents leverage Microsoft 365 Copilot’s advanced natural language processing capabilities combined with direct access to the content of your documents, enabling them to answer questions, summarize information, and extract key data points from a targeted corpus of files.

Unlike generic AI chatbots, these agents are trained and tuned specifically on the documents you designate—such as a folder containing all files related to a litigation matter, contract negotiation, or regulatory compliance. This focused grounding ensures responses are highly accurate, relevant, and tailored to the precise context of your legal work.

“Agents enable you to interact with your case files as if you had a virtual legal assistant who has already read and understood every document in your folder.”

2. Why Agents Are Transformative for Law Firms

The traditional approach to managing case files involves manually searching through folders, opening multiple documents, skimming lengthy depositions, and cross-referencing dates, exhibits, and deadlines. This process is time-consuming, labor-intensive, and prone to human error.

Agents transform this workflow by providing an on-demand AI assistant that has effectively read and internalized every document in your designated case folder. Instead of sifting through dozens or hundreds of files, you simply ask the agent a question in natural language. The agent then synthesizes relevant information from across the entire corpus and delivers accurate, concise answers.

For law firms, this means:

• Increased efficiency: Dramatically reduce the time spent on document review and fact-finding.
• Improved accuracy: Minimize oversight by relying on an AI that cross-checks all available data.
• Enhanced collaboration: Share agents with your legal team to ensure everyone has consistent, up-to-date information.
• Streamlined knowledge management: Convert vast, unstructured case documents into a searchable, interactive knowledge base.

3. Step-by-Step Live Demo Walkthrough

Let’s walk through the process of creating and querying an Agent grounded in a typical case folder in OneDrive. For this example, we will use the folder path /Cases/Acme v Globex/.

a. Navigate to the Case Folder in OneDrive

1. Open your OneDrive interface via the web or desktop app.
2. Locate the primary cases directory, then select the folder named Acme v Globex under /Cases/.
3. Confirm the folder contains relevant files such as pleadings, discovery documents, deposition transcripts, and scheduling orders.

b. Select the Files or Folder

You can create an agent based on a single file, multiple selected files, or an entire folder. For comprehensive coverage, select the entire Acme v Globex folder.

c. Choose “Create Agent” from the Copilot Menu

1. Right-click the selected folder or click the Copilot icon/menu option in the OneDrive toolbar.
2. From the dropdown, select Create Agent.
3. A prompt window will appear to configure your new Agent.

d. Name the Agent and Provide a Description

Choose a clear, descriptive name and provide a brief summary of the matter:

• Agent Name: Acme Case Agent
• Description: “AI assistant trained on all case files related to Acme v Globex litigation, including pleadings, depositions, exhibits, and scheduling orders.”

Click Create to initialize the agent. The system will process and index the documents to ground the agent’s understanding.

e. Query the Agent with Sample Questions

Once created, you can start asking questions directly through the Copilot interface associated with this agent. Try the following sample queries:

“What was the date of the initial complaint?”

“Summarize the key arguments from the plaintiff’s deposition.”

“What are the total damages claimed?”

The agent will return concise, accurate answers by synthesizing information from all relevant documents within the folder.

4. Sample Queries and Expected Outputs

Query	Expected Output
What was the date of the initial complaint?	Specific date extracted from the complaint document (e.g., “January 15, 2024”).
Summarize the key arguments from the plaintiff’s deposition.	A brief, bullet-point summary of major points made by the plaintiff during deposition.
What are the total damages claimed?	Aggregate damages amount as stated in pleadings or damage calculations.
List all exhibits referenced in the complaint.	A list of exhibit numbers and titles with page references.
What discovery deadlines are mentioned in the scheduling order?	Summary of dates and deadlines for discovery phases, including motions and depositions.
Who are the key witnesses identified in the plaintiff’s disclosures?	Names and brief descriptions of witnesses listed in disclosure documents.
Provide a timeline of major case events to date.	Chronological list of filings, hearings, depositions, and orders.
What are the key contract provisions cited in the defendant’s motion to dismiss?	Extract and summarize relevant contract clauses referenced in the motion.

5. Best Practices for Agent Creation

To maximize the usefulness and accuracy of your Agents, follow these best practices:

• Use Descriptive Names: Choose agent names that clearly identify the case and scope (e.g., Acme Case Agent, Globex Discovery Agent).
• Provide Clear Descriptions: Summarize the agent’s purpose and document scope in the description field to aid team members.
• Scope Appropriately: Select files and folders carefully to avoid including irrelevant documents that may confuse the agent.
• Include Comprehensive Documents: Ensure all key pleadings, discovery, correspondence, and orders are included for holistic understanding.
• Regularly Update: Refresh or recreate agents as new case documents are added to maintain accuracy.
• Use Consistent Naming Conventions: This helps when managing multiple agents across related cases or departments.

6. Managing Agents

Updating Agents When New Documents Are Added

Case files are dynamic and continuously updated. To ensure your Agent remains current:

• Add new documents to the same folder or file collection used to create the agent.
• Use the Update Agent option from the Copilot menu to re-index the folder, incorporating recent files.
• If extensive changes occur, consider deleting and recreating the agent for optimal performance.

Sharing Agents with Team Members

Agents can be shared to facilitate team collaboration:

• Agents are tied to the OneDrive folder permissions; ensure team members have access to the underlying files.
• Share the agent link or invite colleagues via the Copilot interface.
• Shared agents enable consistent and synchronized querying across your legal team.

7. Limitations and Workarounds

While Agents are powerful, it is important to understand current limitations and practical workarounds:

• File Type Support: Agents perform best with text-based files such as Microsoft Word, PDFs with selectable text, Excel, and PowerPoint documents. Scanned images or handwritten notes may not be fully processed unless OCR is applied externally.
• Token Limits: Very large case folders with thousands of pages may exceed the AI model’s token limits. To address this:
• Split the case into multiple agents focused on subsets of documents (e.g., pleadings, discovery, billing).
• Regularly archive obsolete files to reduce folder size.
• Contextual Nuance: AI agents interpret language based on available documents but cannot replace attorney judgment. Always review AI outputs critically.
• Update Latency: There may be a delay between adding new documents and the agent reflecting those changes; plan updates accordingly.

8. Advanced Use Cases

To further leverage Agents within your law firm, consider creating multiple agents for different aspects of the same matter. For example:

• Acme Discovery Agent: Focused on discovery documents, interrogatories, and deposition transcripts.
• Acme Billing Agent: Grounded in invoices, payment records, and fee agreements.
• Acme Contract Review Agent: Dedicated to contracts, amendments, and related correspondence.
• Acme Compliance Agent: Monitors regulatory filings, compliance checklists, and risk assessments.

This modular approach enables specialized insights and more granular control over AI-assisted document review across practice areas.

9. Practice Exercise Instructions

For hands-on experience, complete the following practice exercise:

1. Navigate to the sample /Cases/Acme v Globex/ folder in your OneDrive environment.
2. Select the entire folder and create a new Agent named Acme Case Agent with an appropriate description.
3. Once the Agent is created, ask it at least two of the sample questions provided in Section 4.
4. Review the responses carefully and note any insights or clarifications.
5. Optionally, share your agent with a colleague and compare query results.

This exercise will solidify your understanding of how to build and use AI Agents effectively within your legal practice.

Table: Query Type | Example Prompt | Expected Output Type

Query Type	Example Prompt	Expected Output Type
Date Extraction	“What was the date of the initial complaint?”	Specific date (e.g., January 15, 2024)
Summarization	“Summarize the key arguments from the plaintiff’s deposition.”	Concise bullet-point summary
Numeric Aggregation	“What are the total damages claimed?”	Aggregate sum figure
List Extraction	“List all exhibits referenced in the complaint.”	Enumerated list with references
Deadline Identification	“What discovery deadlines are mentioned in the scheduling order?”	List of dates with event descriptions
Witness Identification	“Who are the key witnesses identified in the plaintiff’s disclosures?”	Names and roles of witnesses
Timeline Construction	“Provide a timeline of major case events to date.”	Chronological event list
Contract Clause Extraction	“What are the key contract provisions cited in the defendant’s motion to dismiss?”	Extracted and summarized contract clauses

---

Conclusion

Microsoft 365 Copilot Agents represent a paradigm shift in legal document management and case analysis. By grounding AI assistants directly in your OneDrive case folders, you can unlock powerful, context-aware insights without tedious manual review. Mastering the creation, querying, and management of these agents will streamline your workflows, enhance collaboration, and empower your legal team to focus on higher-value strategic work.

We encourage you to practice creating your own agents and integrating them into your daily casework. In the next lesson, we will explore advanced customization techniques and integrating Agents with other Microsoft 365 tools for seamless end-to-end legal project management.

Thank you for engaging with this lesson. Please proceed to the practice exercise and share your results in the course discussion forum.

Deep Dive: Practical Examples of Case File Agents in Action

To solidify your understanding, let’s examine some real-world scenarios where Case File Agents can streamline your day-to-day legal workflows.

Scenario	Agent Function	Expected Outcome
Automated Document Summarization	Agent scans all pleadings and motions in a case folder and generates concise summaries highlighting key arguments and deadlines.	Attorneys receive a daily briefing email summarizing recent filings, enabling quick case status updates without manual review.
Deadline Tracking and Alerts	Agent parses calendar entries, court orders, and client communications to identify upcoming deadlines and automatically remind responsible team members.	Reduced risk of missed deadlines and improved coordination among legal staff.
Discovery Document Tagging	Agent reviews discovery documents and applies tags based on content type, relevance, or confidentiality level.	Faster filtering and retrieval of discovery materials during case preparation.

Each of these examples demonstrates how integrating intelligent Agents into your case files can reduce manual effort and enhance accuracy.

Common Mistakes When Building Case File Agents

While powerful, deploying Agents effectively requires attention to detail. Here are some pitfalls to avoid:

• Overloading Agents with Tasks: Trying to make a single Agent perform too many functions can reduce performance and complicate query responses. Instead, design specialized Agents for discrete tasks.
• Insufficient Training Data: Agents rely on well-structured data. Poorly organized case files or inconsistent naming conventions limit the Agent’s ability to interpret and act on information.
• Ignoring Security Settings: Remember that case files often contain sensitive information. Ensure Agents and their queries comply with your firm’s data security policies and that proper permissions are set.
• Neglecting Regular Updates: Legal matters evolve rapidly. Failing to update Agents’ configurations to reflect new workflows or document types can render them obsolete.
• Misinterpreting AI-generated Suggestions: AI suggestions should be reviewed critically, especially in legal contexts. Blindly trusting Agents without human oversight can lead to errors.

Troubleshooting Tips for Case File Agents

If your Case File Agents are not functioning as expected, consider the following troubleshooting steps:

Problem	Possible Cause	Recommended Solution
Agent fails to retrieve documents	Incorrect or incomplete folder permissions	Verify and update SharePoint or OneDrive permissions to ensure the Agent has read access.
Agent returns irrelevant results	Poorly defined query parameters or ambiguous instructions	Refine the query language, add specific keywords, or use filters to narrow results.
Agent performance is slow	Agent processing too many files or complex AI tasks simultaneously	Break down tasks into smaller Agents, schedule processing during off-hours, or archive older files.
Agent does not respond to queries	Connectivity or authentication issues with Microsoft 365 services	Check internet connection, re-authenticate your Microsoft account, or restart the Copilot interface.

Advanced Scenarios: Leveraging Agents for Complex Case Management

Beyond basic document retrieval and summarization, advanced users can implement sophisticated workflows that integrate multiple Microsoft 365 services with Case File Agents.

Integration with Microsoft Teams for Real-Time Collaboration

Configure Agents to post automated updates directly to Teams channels dedicated to specific cases. For example, when a new deposition transcript is uploaded, the Agent can notify the litigation team instantly, facilitating rapid review and discussion.

Cross-Referencing Case Files with External Databases

Advanced Agents can be programmed to query external legal databases or court docket systems via APIs, pulling in relevant case law or procedural updates linked to your case files. This helps attorneys stay informed without leaving the OneDrive environment.

Custom Workflow Automation Using Power Automate

Combine Agents with Microsoft Power Automate to trigger multi-step processes, such as:

• Automatically assigning a document for review when uploaded
• Routing client billing information extracted from case files to accounting teams
• Generating follow-up tasks in Microsoft Planner based on Agent-identified deadlines

This layered automation reduces manual task management overhead and prevents missed actions during complex litigation.

FAQ: Common Questions About Case File Agents

Q1: How secure is the data processed by Case File Agents?

A: Case File Agents operate within your Microsoft 365 environment, inheriting the platform’s enterprise-grade security and compliance standards. However, it is crucial to manage access controls carefully and avoid granting Agents permissions beyond what is necessary.

Q2: Can multiple Agents operate on the same case folder simultaneously?

A: Yes. You can deploy multiple Agents specialized on different tasks within the same case folder. Just ensure their responsibilities do not conflict to avoid duplication or inconsistent outcomes.

Q3: How do I update an existing Agent’s behavior or parameters?

A: Access the Agent’s settings in Microsoft 365 Copilot, where you can modify prompts, data sources, or trigger conditions. After saving changes, test the Agent on sample queries before full deployment.

Q4: What types of legal documents can Agents process effectively?

A: Agents perform best with text-based files such as Word documents, PDFs with selectable text, and emails. Scanned images or handwritten notes require OCR processing first to convert to text.

Q5: How can I measure the effectiveness of my Case File Agents?

A: Track metrics like query response accuracy, time saved in document review, and reduction in missed deadlines. Soliciting attorney feedback is also valuable to identify areas for improvement.

Reference Tables for Quick Agent Setup and Query Syntax

Below are handy tables summarizing key commands and best practices for building and querying your Case File Agents.

Agent Setup Step	Recommended Action	Notes
Define Agent Scope	Select specific OneDrive folders or SharePoint libraries	Narrow scope to improve performance and security
Create Prompt Templates	Write clear, task-specific instructions for AI processing	Use plain language and examples to guide responses
Test with Sample Queries	Run a set of typical user questions and verify output accuracy	Iteratively refine prompts based on results
Set User Permissions	Limit Agent access to authorized legal staff	Comply with firm confidentiality policies
Schedule Regular Updates	Review and update Agent configurations quarterly	Adapt to evolving case management needs

Query Syntax	Description	Example
keyword:phrase	Search for documents containing exact phrase	deadline:”motion to dismiss”
AND / OR	Combine multiple search terms	contract AND “payment terms”
NOT	Exclude terms from results	discovery NOT privileged
date:YYYY-MM-DD	Filter documents by date	date:2023-01-15
author:name	Search for documents created by a specific author	author:”John Smith”

Summary and Next Steps

Building and querying Case File Agents unlocks powerful AI capabilities for your law firm, helping you manage case files more efficiently and accurately. As you continue practicing these skills, keep the following in mind:

• Start small—build simple Agents and expand their functions gradually.
• Maintain rigorous data hygiene and security controls.
• Continuously gather feedback from your team to improve Agent effectiveness.
• Explore integration possibilities with other Microsoft 365 tools to automate workflows end-to-end.

In the upcoming lessons, we will explore how to customize Agent interfaces and embed AI-driven insights directly into your document drafting process, further enhancing your firm’s productivity.