Creating Your First Project in ChatGPT
In the previous post, we explored the ChatGPT interface and learned how to use its main features.
In this post, we will create our first ChatGPT Project and learn how Projects can help us organize conversations, files, instructions, and long-term work.
Projects are especially useful when working on a subject over several days or weeks. Instead of keeping everything in separate conversations, you can organize related work inside one Project. ChatGPT can use the conversations and files within the Project as context while you continue working.
What is a ChatGPT Project?
A Project is a workspace for a specific goal or activity.
A Project can contain:
- Conversations
- Files
- Project instructions
- Related context
- Other available tools
For example, you could create Projects for:
- AI Agent Development
- Kubernetes
- .NET Development
- WordPress
- Software Architecture
- Research
- Writing a technical book
Instead of mixing all these subjects in your general ChatGPT conversations, each Project can have its own workspace.
Why Should We Use Projects?
Imagine that you are developing an AI Agent.
During the development process, you may need to discuss:
- Requirements
- Architecture
- Programming
- Database design
- APIs
- Testing
- Documentation
- Security
If all of these discussions are stored in unrelated chats, it can become difficult to find the information you need.
A Project provides a central workspace where related conversations and reference files can be kept together.
This is particularly useful for long-running work because ChatGPT can use relevant context from the Project when continuing your work.
Step 1: Open ChatGPT
Open the ChatGPT Desktop application and sign in to your account.
You can also use ChatGPT through the web application.
After signing in, look at the sidebar on the left side of the screen.
You should see an option for Projects.
The exact appearance of the interface may change as OpenAI updates ChatGPT.
Step 2: Create a New Project
Select Projects and choose the option to create a new Project.
Give your Project a meaningful name.
For this tutorial, let’s create:
AI Agent Development
A descriptive name is important because it helps you immediately understand the purpose of the Project.
Step 3: Add a Project Description
If the Project provides a description field, add a short description explaining its purpose.
For example:
This Project is for learning and developing AI Agents, including architecture, tools, memory, RAG, MCP, APIs, and practical software development.
The description helps you remember what the Project is intended for.
Step 4: Add Project Instructions
One of the most useful features of Projects is Project instructions.
Project instructions tell ChatGPT how you want it to behave inside that Project.
For example, you could add:
Act as a senior software architect and AI Agent development mentor. Explain technical concepts clearly and provide practical examples. Prefer C# and .NET examples when appropriate. When providing code, follow modern development practices and explain important design decisions.
These instructions apply to conversations inside the Project and help keep the responses consistent.
Project instructions are different from the questions you ask in individual conversations. The instructions define how ChatGPT should generally work within the Project.
Step 5: Add Reference Files
You can also add files to a Project.
For example, an AI Agent development Project could contain:
- Architecture documents
- Requirements
- Source code
- API documentation
- PDF documents
- Technical specifications
- Database designs
- Project notes
These files can provide additional context when you ask questions.
For example, you could upload a document called:
AI-Agent-Requirements.pdf
Then ask:
Analyze the requirements in this document and identify the main components required for the AI Agent.
ChatGPT can use the uploaded material as part of the Project context.
Step 6: Create Your First Conversation
Now create a new conversation inside the Project.
For example, start with:
I want to build a customer-support AI Agent for a software company. Help me define the requirements and identify the main components.
ChatGPT can begin working with you on the problem.
You can then continue the same conversation with questions such as:
Now design the architecture.
Then:
What tools should the Agent have?
Then:
How should memory work?
Because these conversations belong to the same Project, they can become part of your ongoing project context.
Step 7: Create Separate Conversations for Different Topics
You do not necessarily need to put everything into one giant conversation.
For example, inside AI Agent Development, you could create separate conversations:
Requirements
Define the requirements for the AI Agent.
Architecture
Design the architecture based on the requirements.
Database
Design the database required by the Agent.
Tools
Define the tools that the Agent should be able to use.
Memory
Design the memory architecture.
Security
Identify security risks and protection mechanisms.
Testing
Create a testing strategy for the AI Agent.
This organization makes the Project much easier to manage.
Step 8: Move an Existing Conversation Into a Project
If you already have a useful conversation outside the Project, you may be able to move it into the Project.
Depending on the current ChatGPT interface, you can use the conversation menu and select Move to project, or drag the conversation onto the Project in the sidebar.
This can be useful when you have already started researching a topic before creating the Project.
Project Memory
Projects have built-in memory and context features.
Depending on your settings and the type of Project, ChatGPT can use information from conversations within that Project to provide more relevant responses.
When creating a new Project, you may also have the option of selecting project-only memory.
With project-only memory, conversations can refer to other conversations within the same Project but not conversations outside that Project. This creates a more self-contained workspace.
This can be particularly useful for a dedicated software development project where you want the context to remain focused.
Example: Our AI Agent Project
Let’s imagine that we create the following Project:
Project Name:
AI Customer Support Agent
Project Instructions:
Act as a senior AI Agent architect and .NET developer. Help design a secure and maintainable customer-support AI Agent. Prefer C# and ASP.NET Core examples. Explain architecture and design decisions clearly. Never assume that an AI-generated solution is correct without verification.
Reference Files:
- Customer requirements
- API documentation
- Database design
- Product documentation
Conversations:
- Requirements
- Architecture
- Database
- AI Model
- Tools
- Memory
- RAG
- Security
- Testing
- Deployment
Now we have a structured workspace for developing the Agent.
Projects Are Not AI Agents
It is important to understand the difference.
A ChatGPT Project is a workspace.
It helps organize:
- Conversations
- Files
- Instructions
- Context
An AI Agent is a software system designed to pursue a goal, use tools, make decisions, and perform actions within defined permissions.
Therefore, creating a Project does not automatically create an AI Agent.
However, Projects can be very useful when learning about and developing AI Agents.
Project vs Normal Chat
A normal ChatGPT conversation might look like this:
User → Question → ChatGPT → Answer
A Project provides a larger working environment:
Project
↓
Instructions
↓
Files
↓
Conversations
↓
Context
↓
ChatGPT
This makes Projects more suitable for long-running work.
A Practical Exercise
Let’s create a Project that we will use throughout this AI Agents Step-by-Step series.
Create a Project called:
AI Agent Development
Add the following Project instruction:
You are my AI Agent development mentor. Help me learn and build AI Agents step by step. Explain concepts clearly and provide practical examples. When programming examples are required, prefer C# and .NET unless another language is specifically requested. Focus on secure, maintainable, production-oriented solutions.
Then create your first conversation:
I want to learn how to build an AI Agent from the ground up. Create a learning plan covering architecture, LLMs, prompting, memory, tools, RAG, planning, security, testing, and deployment.
This Project can now become your personal workspace for the rest of this tutorial series.
What Have We Learned?
In this tutorial, we learned how to:
- Create a ChatGPT Project
- Give the Project a meaningful name
- Add Project instructions
- Upload reference files
- Create conversations inside a Project
- Organize conversations by topic
- Move existing conversations into a Project
- Understand Project memory
- Use a Project as a workspace for AI Agent development
Projects provide an organized environment for long-running work, but they are only one part of the larger AI Agent ecosystem.
In the next post, we will look at Codex, a development-focused tool that can help programmers understand, write, modify, and work with software code.