What are Custom GPTs?
Custom GPTs are AI assistants that users can tailor to perform specific tasks. Using a low-code/no-code interface, users can build Custom GPTs by describing their needs in natural language to GPT Builder.
Creating A Custom GPT
To create a Custom GPT:
- Open ChatGPT, then navigate to the ‘GPTs’ page on the left panel
- Click the ‘+ Create’ button on the upper right corner. This will lead you to the GPT Builder page
- Type in the requirements for your Custom GPT. An example would be “Make a Custom GPT that can summarise the top daily AI trends”
- To further tailor your Custom GPT, click on the ‘Configure’ tab (see the table below for more information)
- Click the ‘Create’ button on the top right corner to start using your Custom GPT. You may also share it with other people.
Understanding the Configure Tab
Under the ‘Configure’ tab are a few customisable settings. Below is a quick guide on what each of them mean.
| Term | What It Does |
| Instructions | Guides the Custom GPT on how it should execute a task |
| Prompt starters | Examples on how a user may prompt the Custom GPT |
| Knowledge | Upload files to provide additional context for your Custom GPT to produce more relevant outputs. Content from the files uploaded to ‘Knowledge’ could appear in the output |
| Custom Actions | Make external APIs available to your Custom GPT, which allows you to connect your CustomGPT to other online tools. This allows it to do more things such as get real-time info, send messages, or interact with apps outside itself. |
Sharing Custom GPTs With Others
On ChatGPT Enterprise, you can place limits on how others use your Custom GPT as a workspace owner. To manage your Custom GPT, click on your profile > ‘Manage workspace’ > ‘GPTs’. Here, you may select who to share your Custom GPT with and control the ways in which users are allowed to make calls to domains.
Referencing Custom GPTs in Any Chat
The @ function allows Custom GPTs to be invoked directly within a regular chat. A list of available GPTs appears when the @ symbol is typed, enabling Custom GPTs to be accessed without navigating away from the ongoing conversation. This keeps specialised capabilities integrated into the workflow and reduces the need for context switching.
Reference: OpenAI
For more learning resources on Custom GPTs, check out the following articles:
