Tutorial 04: Using Custom GPTs for Matter Management (OpenAI)
Learn how to create persistent legal workspaces with Custom GPTs, organize matters, write custom instructions, and manage client confidentiality effectively.
Learning Objectives
By the end of this tutorial, you will:
- Understand how Custom GPTs create persistent legal workspaces
- Set up Custom GPTs for different practice areas and matters
- Write effective custom instructions (your digital playbook)
- Manage client confidentiality with GPT organization
- Organize in-house legal departments using Custom GPTs
- Implement case prioritization and resource allocation
- Conduct matter briefings and team handoffs
- Optimize document organization and naming conventions
Time Required: 45 minutes | Level: Beginner | No technical skills required
Understanding Matter Workspaces
What Are Matter Workspaces in OpenAI?
With ChatGPT, you achieve matter-specific workspaces using:
- Custom GPTs: Persistent instructions and knowledge files per matter or practice area
- Knowledge Files: Documents ChatGPT can reference within a GPT
- Conversation History: Past discussions within each chat thread
Why Matter Workspaces Matter for Legal Work
| Without Workspaces | With Custom GPTs |
|---|---|
| Re-explain context every conversation | Context persists via instructions and knowledge |
| Re-upload documents repeatedly | Documents available in GPT knowledge base |
| Inconsistent advice across sessions | Consistent playbook applied |
| No separation between matters | Clean matter boundaries |
| Risk of cross-contamination | Isolated workspaces per GPT |
Custom GPTs vs. Harvey Vault / Legora Workspace
Harvey Vault: Centralized document storage with AI analysis Legora Workspace: Collaborative environment with saved workflows ChatGPT Custom GPTs: Flexible workspace with full customization control
OpenAI's Advantage: You define exactly how ChatGPT should behave—no rigid templates or workflows imposed by the vendor.
Project Architecture for Law Firms
Recommended Structure
Confidentiality Best Practices
DO:
- Use separate Custom GPTs for each client matter
- Use client codes instead of names if concerned
- Clear or archive GPT when matter closes
- Control who has access to Team/Enterprise GPTs
DON'T:
- Mix multiple clients in one GPT
- Upload documents from different matters together
- Leave sensitive GPTs accessible after matter ends
- Use client names in GPT titles on shared accounts
Creating Your First Matter GPT
Step-by-Step Setup
Step 1: Create the Custom GPT
- Open ChatGPT (chat.openai.com or app)
- Go to Explore GPTs or Create
- Click "Create a GPT"
- Enter name:
[ClientCode] - Matter Description
Step 2: Write Custom Instructions
Custom instructions tell ChatGPT how to behave in this workspace. They're your digital playbook.
Template for Contract Review GPT:
Step 3: Upload Knowledge Base Documents
Add documents ChatGPT should reference:
- Your firm's standard contract templates
- Client's preferred positions document
- Relevant regulatory guidance
- Sample approved agreements for reference
Step 4: Test the GPT
Start a conversation and verify ChatGPT follows your instructions:
Custom Instructions Templates
Template: Litigation Support GPT
Template: Legal Research GPT
Template: Transaction Support GPT
In-House Legal Department Organization
Setting Up GPTs for Corporate Legal Teams
For in-house counsel, organize Custom GPTs by function:
Matter Intake Workflow
Create a Matter Intake GPT with these custom instructions:
Business Partner Collaboration Framework
Add this guidance to relevant Custom GPTs:
Board Reporting Template
Create a Board Reporting GPT with this structure:
Case Prioritization & Resource Allocation
AI-Recommended Matter Priority Framework
Create a Matter Prioritization GPT with these instructions:
Resource Allocation Workflow
Use this prompt in your prioritization GPT:
Deadline Management System
Create a Deadline Tracker GPT:
Briefing & Catch-Up Workflows
Case Briefing on Return from Absence
Create a Matter Briefing GPT for this use case:
Matter Overview Generation
Use this prompt for any Custom GPT:
Status Summary Prompts
Create quick-reference briefing prompts:
Team Handoff Documentation
Use this template when transitioning a matter:
Document Organization & Naming Conventions
AI-Suggested Naming Conventions
Create a Document Organization GPT with this guidance:
Folder Structure Optimization
Transaction Folder Structure:
Litigation Folder Structure:
File Cleanup Workflows
Regular Cleanup Schedule:
Weekly:
- Delete working drafts (keep final versions only)
- Remove duplicate files
- Archive superseded versions
Monthly:
- Review and organize new documents
- Update folder structure if needed
- Remove access for departing team members
- Archive closed items
Quarterly:
- Full audit of matter folders
- Identify and remove obsolete materials
- Update document indexes
- Verify naming consistency
Cleanup Prompts for ChatGPT:
Version Control Best Practices
Best Practices & Practical Exercises
Exercise 1: Set Up Your First GPT with Prioritization
Task: Create a Custom GPT for managing 3-5 active matters
Instructions:
- List your active matters
- Create one master GPT for matter management
- Add custom instructions including:
- Prioritization matrix from above
- Deadline tracking requirements
- Resource allocation framework
- Test with prompt: "Rank my current matters by priority: [list matters]. Consider deadlines, business impact, and complexity."
Exercise 2: Create a Matter Briefing Template
Task: Build a catch-up brief for a real matter
Instructions:
- Select a matter you recently returned to
- Create a dedicated Briefing GPT
- Upload key documents to knowledge base
- Use the "Matter Briefing" prompt template
- Generate comprehensive catch-up brief
- Review for completeness and accuracy
Exercise 3: Organize Documents Using Naming Conventions
Task: Audit and reorganize a matter file
Instructions:
- Select a matter with multiple documents
- Create a list of current file names
- Propose new names using the framework above
- Create folder structure (transaction or litigation template)
- In ChatGPT, ask: "Help me organize these files according to [type] best practices. Here are my current files: [list]. Suggest naming and folder structure."
- Implement the recommended structure
Exercise 4: Design In-House Legal Department Organization
Task: Create Custom GPTs matching your corporate structure
Instructions (for in-house counsel):
- Map your legal department functions
- Create Custom GPTs for each major function (governance, operations, compliance, IP, etc.)
- For each GPT, write custom instructions including:
- Matter intake procedures
- Business partner collaboration guidelines
- Board reporting requirements
- Test with intake prompt: "New matter just came in: [description]. How should I handle this?"
Exercise 5: Implement Priority & Resource Allocation
Task: Set up your prioritization system
Instructions:
- List all current open matters
- Create Priority GPT with instructions from above
- Upload matter list to knowledge base
- Add attorney availability information
- Ask ChatGPT: "Please prioritize these matters and recommend resource allocation. Consider deadlines, complexity, business impact, and available capacity."
- Review recommendations and adjust as needed
Managing GPTs Over Time
Updating Custom Instructions
As matters evolve, update your instructions:
- Add newly agreed terms as "acceptable"
- Update case status and key facts
- Add lessons learned from negotiations
- Incorporate new team members' preferences
GPT Lifecycle
Phase 1: Setup (Matter opening)
- Create GPT with detailed instructions
- Upload initial documents
- Test with calibration questions
Phase 2: Active Use (Matter ongoing)
- Regular conversations within GPT
- Upload new documents as received
- Refine instructions based on feedback
- Update deadline and priority tracking
- Conduct periodic catch-up briefings
Phase 3: Archive (Matter closing)
- Export any needed conversation logs
- Document final outcomes in instructions
- Remove or redact sensitive materials
- Consider retention/deletion policy
- Archive in matter management system
Phase 4: Reuse (Templates)
- Create template from successful GPT
- Strip client-specific information
- Use as starting point for similar matters
Team Collaboration with ChatGPT
ChatGPT Team Features
With ChatGPT Team or Enterprise:
- Share Custom GPTs across team members
- Consistent instructions for practice groups
- Admin controls for access management
- Audit capabilities
Team plan pricing is set by OpenAI and may change. Verify current pricing on the official ChatGPT site.
Collaboration Best Practices
Firm-Wide Playbooks:
- Create "template" GPTs for common work types
- Partners/senior associates define standards
- Associates use pre-configured GPTs
- Ensures consistency across the team
Matter-Specific GPTs:
- Lead attorney creates and owns GPT
- Team members given access as needed
- Single source of truth for matter context
- Clear handoff when staffing changes
Comparing to Enterprise Features
Custom GPTs vs. Harvey Vault
| Feature | ChatGPT Custom GPTs | Harvey Vault |
|---|---|---|
| Document storage | In-GPT knowledge base | Centralized vault |
| Custom instructions | Fully customizable | Pre-built workflows |
| Search across docs | Within GPT | Enterprise-wide |
| Setup time | Minutes | Days (implementation) |
| Customization | Unlimited | Limited to Harvey options |
When Enterprise Tools Make Sense
Choose Harvey/Legora when you need:
- Very large-scale bulk processing and cross-matter automation
- Enterprise SSO and compliance
- Cross-matter search and analytics
- Vendor management of AI infrastructure
- Pre-built integrations with your DMS
Choose Custom GPTs when you need:
- Full control over AI behavior
- Cost-effective solution for smaller teams
- Rapid setup without IT involvement
- Flexibility to experiment with workflows
- Integration with your custom tools via API/function calling
Homework Before Tutorial 05
-
Create 5 Custom GPTs:
- One for a current matter
- One for matter prioritization/intake
- One for a practice area playbook
- One for legal research in your specialty
- One for document organization (if managing multiple matters)
-
Write detailed custom instructions for each
-
Upload 5+ documents to at least two GPT knowledge bases
-
Test each GPT with real work tasks:
- Matter prioritization analysis
- Catch-up briefing
- Document organization recommendations
- Deadline tracking
-
Iterate on instructions based on output quality
-
Document your naming conventions and folder structures for future reference
Quick Reference: Custom Instruction Elements
Do This Now
- Create a Custom GPT for one client matter with custom instructions
- Upload key documents to the knowledge base
- Test that ChatGPT follows your instructions before using on real work
Continue to Playbooks -> | Previous: Document Analysis
Related family pages
- Claude Projects - Same concepts with Claude
- Core Concepts - Platform-neutral legal workflow model