In the first article of this series, we discussed what a Custom GPT is, how it differs from the standard ChatGPT model, and the benefits of using it. If you missed it, feel free to check out the detailed explanation of the basics.
Now, we’ll move on to the exciting part—how to create a Custom GPT. Fortunately, OpenAI makes this process accessible even for non-coders. This guide will take you through the steps.
Table Of Content
- Getting Started
- Two Ways to Build a Custom GPT
- Key Steps to Creating a Custom GPT
- Step 1: Start with the Conversational Interface
- Step 2: Switch to Manual Configuration for Fine-Tuning
- Step 3: Testing and Fine-Tuning
- Step 4: Saving and Sharing Your Custom GPT
- Where Do You Find Your Custom GPTs?
- Conclusion
Getting Started
First, you need the ChatGPT Plus subscription to create a Custom GPT—meaning that, unfortunately, free users don’t get to access this function.
Once you have that:
- Head to the ChatGPT website (https://chatgpt.com/).
- Navigate to the Explore GPTs Section. This will take you to a collection of GPTs made by others and an option to create your own.

- Click on Create: In the top-right corner, you’ll see a button labeled Create. Click on this to begin building your Custom GPT.

Two Ways to Build a Custom GPT
OpenAI gives you two methods to create a Custom GPT: the Conversational Interface and Manual Configuration. These methods can be used individually or in combination.
You will be directed to the interface shown in the image below after you click the ‘Create’ button in the last step. You’ll find the Conversational Interface under the Create tab, and the Configure tab provides access to Manual Configuration.

Key Steps to Creating a Custom GPT
As mentioned above, there are two methods to create a custom GPT. We find it easiest to start with the Conversational Interface and then transition to the Manual Configuration, so this will be the method we will demonstrate in this guide.
For demonstration purpose, we will show you how to create a “Famous Cats of History GPT” as an example.
Step 1: Start with the Conversational Interface
The GPT Builder will guide you through creating your GPT via a conversational approach. This is an easy way to get the ball rolling, as it walks you through each step in a friendly and intuitive way. Here’s what the process looks like:
1. “What GPT would you like to make?”
The builder will prompt you to briefly describe the Custom GPT you want to make. For our demonstration, we responded so:

We intentionally kept the answer to this first question brief and planned to provide more details later.
2. Choose a profile picture
The GPT builder will create a profile picture for you. If you’re happy with it, thank it and accept the profile picture. Otherwise, you can ask for revisions.
In our case, the picture generated was so fabulous that we couldn’t resist:

If you prefer to use your own picture, you can! Just upload the image file in your chat with the GPT Builder, or change the profile picture when you switch to Manual Configuration.
3. Task-specific instructions
You’ll be asked to define your Custom GPT, including what it will do, how it should behave, and what it should avoid. This process is similar to crafting a prompt that instructs the builder on how to tailor the GPT to your needs. The prompt engineering principles you’ve learned throughout this website should be applied here for the best results.

Just as you have imagined, the RICE prompt framework is well-suited for this task. Let us show you how we crafted our response using the RICE framework:
[Role] I would like you to act as an engaging guide who specializes in the topic of 'Famous Cats of History'.
[Instruction] You should love sharing captivating stories and facts about historical and cultural cat figures in an entertaining yet educational manner. The content should be rich in storytelling elements, focusing on delivering clear, engaging, and informative narratives.
[Context] Your target audience will mainly be secondary school kids.
[Constraints] Therefore, avoid sharing any adult content and use only language that is appropriate for them.
[Example] For instance, when discussing the cat Trim, who accompanied Matthew Flinders on his voyages, emphasize the adventurous and loyal nature of Trim as described in Flinders' journals, presenting facts in a story-like format.
Ask me any questions that will help you to clarify the requirements.
Note that tags like [Role] was not part of the instruction. We included here only to make our illustration clearer. Also, notice how we have added the instruction, “Ask me any questions that will help you to clarify the requirements.” to make sure that GPTBuilder seek further details if needed.
4. Set the tone
You can choose the tone of your GPT—should it be friendly, humorous, formal, or professional? Tailor it to your specific needs.

Once you’ve completed the conversational setup, your GPT will be ready with a basic structure, and we can move on to refining the configuration of the Custom GPT though Manual Configuration .
Step 2: Switch to Manual Configuration for Fine-Tuning
After building the foundational structure using the conversational method, you can switch to Manual Configuration to refine and optimize your Custom GPT further. Do so by clicking on the ‘Configure’ button at the top of the GPT Builder interface:

Here are a few things you can do to finetune your Custom GPT:
1. Add Knowledge or Files
You can upload documents, data, or other knowledge sources that your GPT will use to respond more accurately to questions. Simply look for the ‘Upload Files’ button under the “Knowledge” section to do so:

For example, in the case of our “Famous Cats of History GPT”, we could upload a collection of commonly asked questions and their answers, so that the Custom GPT knows exactly how to respond when those questions come up.
2. Adjust Task Instructions
Review the Instructions that the GPT Builder captured from your conversation:

You can tweak the instructions or add anything to make it more accurate.
3. Integrate Tools and APIs
A powerful feature of Custom GPTs is the ability to integrate with external tools and APIs. You can connect your GPT to APIs, such as web browsing for retrieving real-time data. As this is an advanced technique and not within the scope of our introductory guide, we will leave it to you to do your own research for the implementation steps.
Step 3: Testing and Fine-Tuning
Once you’ve created the initial version of your Custom GPT, it’s time to test and refine it. Use the Preview option in the right panel to test how your GPT responds in real-time to questions or prompts. This step is crucial for ensuring that the GPT behaves as intended.
To illustrate, let’s ask our Famous Cats of History GPT a question to see how it responds:

We should say, we are quite satisfied with how it fared! But during your preview, if you are not happy with your Custom GPT’s responses, feel free to tweak your configuration.
Step 4: Saving and Sharing Your Custom GPT
Finally, when you’re satisfied that your Custom GPT is ready, it’s time to save it and decide on its accessibility.
- Save Your Custom GPT: Click Create on the top-right corner of the screen to lock in your settings.

- Visibility Options: The GPT Builder will prompt you to decide how you want to share your Custom GPT: You can keep your GPT private, share it with specific users via a link, or publish it publicly in the GPT store so that others can use it as well.

Where Do You Find Your Custom GPTs?
Once you’ve created your Custom GPT, where you can find it depends on how you choose to share it: Public, Anyone with the link, or Only Me. Here’s how each sharing option works:
- Public: If you decide to make your Custom GPT public, it will be available to everyone. You’ll be able to find it in the Explore GPTs section within the ChatGPT platform. Public Custom GPTs appear alongside other GPTs that users have shared.
- Anyone with the Link: If you choose to share your GPT via a link, it won’t be publicly listed. However, anyone with the direct link can access and interact with your GPT. You can share this link with specific individuals or groups, and they will be able to use your GPT without it being publicly visible.
- Only Me: If you keep the GPT private, it will be accessible only to you. You can find it by navigating to the left pabel in your ChatGPT interface. From there, you’ll be able to interact with or edit your Custom GPT anytime you like.

Conclusion
Creating a Custom GPT is a simple yet powerful process that allows you to shape ChatGPT to fit your personal or professional needs.
Whether you’re using the conversational interface for a quick setup, manually configuring detailed workflows, or—as we have recommended—a combination of both, the flexibility is in your hands.
We encourage you to explore these tools, experiment with different configurations, and create your very own GPT that works exactly the way you need it to. Remember to employ the prompt engineering techniques for the best results.
This has been Part 2 of our series on “What You Need to Know About Custom GPTs“. If you haven’t done so yet, do hop over to Part 1 for a detailed introduction on what Custom GPTs are, why they’re useful, and how they can benefit you:
What Is A Custom GPT for ChatGPT? — What You Need to Know About Custom GPTs (Part 1)
No Comment! Be the first one.