Hey everyone,
I’m looking for a way to use a custom GPT (with my specific instructions) inside a project or folder in ChatGPT. Right now, it looks like I can either organize chats into projects or use a custom GPT, but I can’t combine both. Has anyone found a workaround for having a custom GPT work directly within a project folder so I can have both organization and my custom instructions together?
Thanks in advance!
Ola, so there are multiple ways on how to make it work ill mention it below also I attached some screenshot of one of my project folder for more clarity
Option A: • Paste your custom instructions into the project
Copy your full custom gpt instructions the stuff that defines its personality, rules, goals, and any background info you want it to always have.
Go to your project folder in ChatGPT, click the Instructions button at the top right, and paste those instructions in.
Naming it is would make it easier it easier adopt the custom instruction
That way, even if just opens a new chat and types “hi (custom gpt name),” it instantly knows it’s supposed to act as (custom gpt ) and follow your custom workflow.
every new instance in this project will pick up your identity, rules, and style even if the chat history is blank.
Option B: Upload Reference Files to the Project
You can upload your custom instruction (as PDF, DOCX, TXT ) and just start new chat and either you can just tell the instance to analyze the documents or it can do it by itself (mostly GPT 5 models)
If you ever need the custom gpt to reference something, just tell it to “check the uploaded file,” and it can pull info or context directly from what you uploaded.
This is perfect if you have complex prompts, style guides, or examples you want to reuse, or if you need to share key info with anyone else working in your project.
I do this myself it keeps everything organized in one place, and saves time since you don’t have to search for stuff or rewrite instructions over and over.
Option C: Move Existing Chats Into the Project and Reset the instance
If you’ve got chats you started before, you can move them right into your project folder just use the three-dot menu next to the chat and pick “Move to project.”
Once it’s inside the project, you can switch the instance if you want (like from GPT-4o to your custom gpt , or even just back and forth, in the chat) also make sure the custom instruction in the instruction box
From then on, the new instance in will adopt your custom instruction but its always better to use the custom gpt name
This is handy if you already have important threads going, and you want them to start using your project’s logic and memory without starting from scratch.
Some extra information that might be useful:
Use “Project-Only” Memory When Creating Your Project Folder
When you first create your project folder, you will see 2 option for memory ( default, project only)
Project Only:
This keeps everything in your project totally separate. The custom gpt will have access to the memories (chat sessions) within this folder and wont reference or recall anything outside of it
Default memory:
With this mode, your custom gpt inside the project folder can access memories of chats, files outside the project folder
I always go with Project-Only memory with i want to create custom gpt that have different way of instruction (automation, mini agents etc) because it keeps my workflow
Hope that helps , let me know if you need further help : )
