Feature Idea
Allow users to invoke the behavior and instruction set of any of their Custom GPTs from any chat, by using a simple @CustomGPTName mention. When a user types @GrammarGPT (or any other custom GPT) in a message, ChatGPT should process that specific input using the instructions and logic of the selected custom GPT, returning a response as if that GPT handled the message, even if the user is not inside that GPT’s dedicated chat window.
Why It’s Useful
- Increases flexibility: Users can access different skills and specialized behaviors without switching conversations or copying instructions manually.
- Saves time: Reduces friction for power users who rely on multiple GPTs (e.g., grammar correction, code formatting, creative writing, etc.).
- Better workflow: Enables complex, multi-tool workflows inside a single chat, increasing productivity and user satisfaction.
- Reduces confusion: Users expect that selecting a GPT with
@would “activate” its configuration. Current implementation is just an autocomplete for sharing, not actual invocation.
How It Could Work
- Mention detection: When a message starts with
@GPTName(or even inline), the system identifies if it matches a custom GPT owned by the user (or a public GPT). - Context switch: For that message only, process the input using the instructions and configuration of the mentioned GPT.
- Response format: The answer is clearly marked as being “from” the invoked GPT, possibly showing its avatar or name for clarity.
- Multiple mentions: If several are mentioned, the user could select which one to apply, or the system could reply with each output (advanced use).
Example
User types in any chat:
@GrammarGPT Please make this paragraph more formal.
Expected behavior: The response is generated using the logic/instructions of “GrammarGPT” regardless of the current chat context.
Thank you for considering this feature!