I have a question about embedding thread

Hello, I have a question about embedding.

I would like to use the currently embedded assistant as an API.

Here’s what I’m curious about:

  • When trying to use the embedded assistant, you cannot receive a single json like chatgpt api. Should I do it in a threaded way?
    If you don’t need to use it, please provide guidance.

The purpose is
Give limited criteria to gpt in advance,
I want to ask questions every time so that you can choose the appropriate one within the criteria given in advance.

Instead of using a thread, ask and receive questions like the chatgpt API!

Translated using a translator.
The question may be strange.

thank you