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're 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've set—no need to re-explain anything.

Why Use Projects?

In a standard ChatGPT conversation, context resets with each new chat. That works fine for one-off questions, but it becomes inefficient when you're working on something over days or weeks. You end up re-uploading the same files, re-explaining your goals, and losing track of which conversation had the output you need.

Projects solve this by providing persistent context across conversations, a central place for reference files, custom instructions that apply automatically to every chat in the project, and a clean organizational structure that keeps related work together.

How to Create a Project

Step 1: Start a New Project

In the ChatGPT sidebar, click the Projects section (or the + icon next to it). Give your Project a clear, descriptive name—something like "Fall 2026 Course Redesign," "Annual Report Drafting," or "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. This is where you tell ChatGPT what the project is about, what role it should play, and any constraints or formatting preferences.

For example, an instructions block for a grant writing project might look like:

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 that will be available in all conversations within it. Supported file types include PDFs, Word documents, spreadsheets, text files, images, and more.

Examples of useful Project files include style guides or templates, policy documents or handbooks, datasets you're actively working with, prior drafts or versions of a document, and reference materials like research papers or specifications.

Files uploaded at the Project level are distinct from files uploaded in a single conversation. Project-level files persist and are accessible in every chat within the Project. Files uploaded in a specific conversation are only available in that conversation.

Step 4: Start Conversations

Once your Project is set up, create new conversations inside it as needed. Each conversation inherits the Project's instructions and has access to the uploaded files. You might have separate conversations for different aspects of the work—one for outlining, one for data analysis, one for revision—while all of them share the same foundational context.

Practical Tips

Name your conversations. As your Project grows, you'll accumulate multiple conversations. Give each one a clear name so you can find specific outputs later. ChatGPT auto-generates titles, but renaming them to something descriptive helps significantly.

Keep instructions focused. Project instructions work best when they're specific to the work at hand. Avoid making them too broad or loading them with conflicting guidance. If you're working on two very different types of tasks, consider using two separate Projects.

Update files as your work evolves. If a reference document changes—a new draft, updated data, revised policies—swap out the old file for the new one. ChatGPT will use whatever is currently uploaded, so keeping files current ensures your responses stay accurate.

Use Projects for recurring workflows. Projects aren't just for one-time initiatives. If you have a recurring task—like weekly reporting, monthly data review, or ongoing documentation work—a standing Project keeps your context intact over time.

Don't overload a single Project. While it's tempting to put everything into one workspace, Projects work best when they're scoped to a coherent body of work. If a Project starts covering too many unrelated topics, the instructions and files become diluted and less effective.

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, and accessibility guidelines. Project instructions tell ChatGPT to focus on active learning strategies and backward design principles. Over several conversations, they develop new learning objectives, draft assignment descriptions, and build a revised course schedule—all within the same context.

Research Initiative A researcher sets up a Project for a multi-month study. They upload their literature review notes, data collection instruments, and preliminary datasets. Separate conversations handle different phases: one for refining the methodology section, one for running exploratory data analysis, and one for drafting the results narrative. The Project's instructions ensure a consistent academic tone and remind ChatGPT to flag any statistical claims that need verification.

Departmental Reporting An administrative team creates a Project for their annual report. They upload last year's report, this year's data summaries, and the provost's reporting template. Instructions specify the required format, section headings, and word count targets. Team members (each in their own account) can create similar Projects using the same files to work on their respective sections.

IT Documentation A technical team creates a Project for maintaining a set of runbooks and knowledge base articles. They upload their documentation style guide and existing articles as examples. When a new system is deployed or a process changes, they start a new conversation within the Project to draft updated documentation that automatically follows the team's standards.

Event Planning A staff member creates a Project for a campus conference. They upload the venue contract, prior year's schedule, speaker bios, and budget spreadsheet. Across multiple conversations, they draft the event agenda, write speaker introduction scripts, create attendee communications, and analyze budget scenarios—all with consistent awareness of the event details.

Projects vs. Custom GPTs

Projects and Custom GPTs serve related but different purposes, and understanding the distinction helps you choose the right tool.

Projects are personal workspaces. They are best for ongoing, multi-conversation work where you need persistent files and instructions. They are private to your account and cannot be shared with others. They support ChatGPT's built-in capabilities like web browsing, data analysis, and image generation.

Custom GPTs are shareable, purpose-built assistants. They are best for creating a reusable tool that others can access. They can be configured with specific instructions, knowledge files, and enabled capabilities. They can be shared via link with other ChatGPT Edu users in your institution.

In short: use a Project when the work is yours and ongoing. Use a Custom GPT when you're building a tool for others to use. In many cases, you might use both—developing something in a Project and then packaging the results into a Custom GPT for broader use.

Limitations

• Projects are not shared. There is no way to give another user access to your Project. If multiple people need the same context, each person would need to set up their own Project with the same files and instructions, or you can build a Custom GPT to share the core functionality.
• File size and quantity limits apply. Very large files or a large number of uploads may hit storage limits. Prioritize the most essential reference material.
• Instructions have a character limit. Keep your Project instructions focused and concise. If you find yourself writing pages of instructions, consider whether a Custom GPT with more structured configuration might be a better fit.
• Context windows are still finite. Even with Project-level files and instructions, ChatGPT has a limit on how much information it can hold in memory within a single conversation. Very long conversations may start to lose earlier context.