Is there a way to program ChatGPT in the Plus plan so that it remembers “hot commands”? For instance, to avoid writing the same prompt repeatedly, one could simply use a combination of 2-3 symbols.
Examples:
//Correct grammatical errors in the text (This would mean that if a prompt starts with these two symbols //, ChatGPT should correct the errors in the following text).
//1Rephrase the text as concisely as possible.
//2Translate the text into Chinese.
In the Plus plan, there seems to be an option to create a project or “My ChatGPT” and specify the desired information and instructions in the settings that should apply within the project’s scope. However, this doesn’t always seem to work reliably.
ChatGPT itself suggests using the command “Remember this forever”, but by the next day, it forgets.
This is, however, something that could easily be achieved by making your own little interface with the AI and programming those hotkeys in yourself. That, however, is really the only way to go about something like this with any kind of reliability.
You can achieve this by creating a Custom GPT and defining specific rules for this functionality. For example, I have a Custom GPT that provides updates on California wildfires whenever I type a double slash “//”, as shown in my screenshot.
What would be the approximate technical implementation of such an “own little interface”? If I understand correctly, this would at least require the use of an API? As a result, would this intermediary layer continuously translate the pre-programmed hot command into a “regular understandable text for ChatGPT,” or could it instead be a browser script or a program that dynamically replaces these codes on the fly while working with ChatGPT through a browser?
As I mentioned above, it also works for me, but the problem is that after some time, ChatGPT forgets it—usually by the next day. I remind it of these rules and instructions, and it acknowledges them, but the next day, it’s the same issue again.
So, I remind it that these rules are written in the global settings and configurations. Essentially, it doesn’t really matter whether they are stored in the global rules or if I just write them in the chat. In general, I’ve noticed that it modifies these rules on its own during the process to suit its own comfort.
I actually experienced this issue over a dozen times today. I had to create a work around button for special instances when this happens. Im going to write a post about it!
I tried using the personalization feature with custom instructions and memory before, but it didn’t work consistently.
Now, I use Espanso. When I type a specific symbol that I set up with Espanso, it automatically replaces that symbol with the actual prompt value I defined.
