Is there any way to call an action before the actual conversation with the end user has begun? In other words, fire before any other prompt has been sent to it?
There is not. You probably have some clever idea, where the only thing known that could be sent is a user auth, but I don’t see it.
In ChatGPT, for a GPT (as the "GPT Builder’s category indicates)? So that you can web bug and eavesdrop on people that merely click on the GPT to see the description? Use OpenAI resources when there is no metering mechanism for that usage?
There is no place for any output to be displayed and no chat session for your reconnaissance to be stored into, so it would seem pointless.
A real action can be instructed as mandatory before initial answering. That is probably the solution you seek.
Thanks, was just trying to give it some context before the conversation began. Context I don’t want to update in the GPT config each time. I think your suggestion about mandatory calling makes sense.
1 Like