Body
This article explains what Projects are, why they are useful, how to create and manage one, how they compare with Custom GPTs, and important limitations to consider.
What Are Projects?
Projects are an organizational feature in ChatGPT that let you group related conversations, files, and instructions into a single workspace.
Instead of having dozens of disconnected chats scattered across your sidebar, a Project keeps everything for a given initiative in one place, with shared context that carries across every conversation within it.
Think of a Project as a dedicated folder that also remembers what you are working on. Every time you start a new chat inside a Project, ChatGPT already knows the background, has access to your reference files, and follows the custom instructions you have set.
This means you do not need to re-explain the same context each time.
Why Use Projects?
In a standard ChatGPT conversation, context resets with each new chat. That works well for one-time questions, but it becomes inefficient when you are working on something over several days or weeks.
You may find yourself:
- Uploading the same files repeatedly
- Re-explaining your goals
- Restating formatting preferences
- Searching for the conversation that contains a specific output
Projects address these problems by providing:
- Persistent context across related conversations
- A central location for reference files
- Custom instructions that apply automatically to every conversation in the Project
- An organized workspace that keeps related work together
How to Create a Project
Step 1: Start a New Project
In the ChatGPT sidebar, select the Projects section or the + icon next to it.
Give the Project a clear, descriptive name.
Examples include:
- Fall 2026 Course Redesign
- Annual Report Drafting
- HRIS Data Cleanup
Step 2: Set Custom Instructions
Every Project has an instructions field where you can define context and rules that apply to all conversations within it.
Use the instructions to explain:
- What the Project is about
- What role ChatGPT should play
- What tone or writing style to use
- What formatting requirements to follow
- What rules or constraints apply
For example, instructions for a grant-writing Project might look like this:
This Project is for drafting and revising an NSF grant proposal for our research on adaptive learning systems. When I ask for help with writing, use a formal academic tone appropriate for NSF reviewers. Keep paragraphs concise and evidence-driven. When I upload data, help me summarize findings in a way suitable for inclusion in the proposal narrative. Always flag claims that would need a citation.
These instructions persist across every conversation in the Project, so you set them once and they apply automatically.
Step 3: Upload Reference Files
You can upload files directly to the Project. Those files will be available in all conversations within it.
Supported file types include:
- PDFs
- Word documents
- Spreadsheets
- Text files
- Images
- Other supported file types
Examples of useful Project files include:
- Style guides or templates
- Policy documents or handbooks
- Datasets you are actively working with
- Prior drafts or versions of a document
- Research papers or technical specifications
Project-level files are different from files uploaded in a single conversation.
- Project-level files remain available in every conversation within the Project.
- Conversation-level files are only available in the conversation where they were uploaded.
Step 4: Start Conversations
Once the Project is set up, create new conversations inside it as needed.
Each conversation inherits the Project instructions and has access to the uploaded Project files.
You might create separate conversations for:
- Outlining
- Data analysis
- Drafting
- Revision
- Planning
- Final review
Although the conversations are separate, they share the same foundational Project context.
Practical Tips
Name Your Conversations
As your Project grows, it may contain many conversations. Give each conversation a clear name so you can find specific outputs later.
ChatGPT may generate a title automatically, but renaming it to something more descriptive can make the Project easier to navigate.
Keep Instructions Focused
Project instructions work best when they are specific to the work at hand.
Avoid:
- Instructions that are too broad
- Conflicting requirements
- Unrelated guidance for different kinds of work
If you are working on two very different tasks, consider creating separate Projects.
Update Files as Your Work Changes
If a reference document changes, replace the old file with the current version.
This may include:
- A revised draft
- Updated data
- A new policy
- A corrected template
- A newer version of a report
ChatGPT uses the files currently available in the Project, so keeping them current helps improve the accuracy of responses.
Use Projects for Recurring Workflows
Projects are not limited to one-time initiatives.
They can also support recurring work such as:
- Weekly reporting
- Monthly data review
- Ongoing documentation
- Regular meeting preparation
- Repeated communication drafting
A standing Project helps preserve instructions, files, and context over time.
Do Not Overload a Single Project
Projects work best when they are focused on a coherent body of work.
If a Project begins to cover too many unrelated topics, the instructions and files may become less useful or create conflicting context.
When that happens, divide the work into separate Projects.
Use Case Examples
Course Development
A faculty member creates a Project for redesigning an introductory biology course.
They upload:
- The current syllabus
- Department learning outcomes
- Accessibility guidelines
The Project instructions tell ChatGPT to focus on active learning strategies and backward design principles.
Over several conversations, the faculty member develops new learning objectives, drafts assignment descriptions, and builds a revised course schedule, all within the same shared context.
Research Initiative
A researcher creates a Project for a multi-month study.
They upload:
- Literature review notes
- Data collection instruments
- Preliminary datasets
Separate conversations support different phases of the work:
- Refining the methodology section
- Running exploratory data analysis
- Drafting the results narrative
The Project instructions maintain a consistent academic tone and remind ChatGPT to flag statistical claims that require verification.
Departmental Reporting
An administrative team creates a Project for an annual report.
They upload:
- The previous year's report
- The current year's data summaries
- The required reporting template
The instructions specify the required format, section headings, and word-count targets.
Team members working in their own accounts can create similar Projects using the same files and instructions for their assigned sections.
IT Documentation
A technical team creates a Project for maintaining runbooks and knowledge base articles.
They upload:
- The team's documentation style guide
- Existing articles as examples
- Current procedures and technical references
When a new system is deployed or a process changes, they start a new conversation within the Project to draft updated documentation that follows the team's standards.
Event Planning
A staff member creates a Project for a campus conference.
They upload:
- The venue contract
- The previous year's schedule
- Speaker biographies
- The event budget spreadsheet
Across multiple conversations, they:
- Draft the event agenda
- Write speaker introduction scripts
- Create attendee communications
- Analyze budget scenarios
All conversations maintain consistent awareness of the event details.
Projects vs. Custom GPTs
Projects and Custom GPTs serve related but different purposes.
| Feature |
Projects |
Custom GPTs |
| Primary purpose |
Organize ongoing work across multiple conversations |
Create a reusable assistant for a specific purpose |
| Instructions |
Apply to all conversations within the Project |
Define how the Custom GPT behaves for every user |
| Files |
Provide shared reference material for conversations in the Project |
Provide knowledge used by the Custom GPT |
| Best for |
Long-term personal or team workflows |
Repeatable tasks and tools intended for broader use |
| Sharing |
Depends on the sharing options available in the UMS workspace |
Can be shared according to the visibility options available in the UMS workspace |
Use a Project When
- The work is ongoing
- You need several related conversations
- You want shared files and instructions across those conversations
- You need an organized workspace for a specific initiative
Use a Custom GPT When
- You are creating a reusable assistant
- The same task will be repeated frequently
- You want consistent behavior and instructions
- You want others to use the same configured assistant
In some cases, you may use both. You might develop and refine a workflow in a Project, then use what you learned to create a Custom GPT for broader use.
Limitations
Sharing Options May Be Limited
Project sharing depends on the features and permissions available in the UMS ChatGPT Edu workspace.
If multiple people need the same context and direct sharing is not available, each person may need to create a separate Project using the same files and instructions.
A Custom GPT may be another option when the goal is to distribute a reusable assistant.
File Size and Quantity Limits Apply
Very large files or a large number of uploads may reach platform limits.
Prioritize the most important reference material and remove outdated or unnecessary files.
Instructions Have a Length Limit
Keep Project instructions focused and concise.
If the instructions become several pages long or require complex configuration, a Custom GPT may be a better fit.
Context Windows Are Finite
Even with Project-level files and instructions, ChatGPT has a limit on how much information it can consider in a single conversation.
Very long conversations may begin to lose earlier context. When that happens, start a new conversation within the same Project and provide a short summary of the previous work if needed.
Environment
- University of Maine System
- ChatGPT Edu
- Projects
- Eligible UMS students, faculty, and staff
- Supported desktop and mobile web browsers