How do I use embeddings so gpt-3-5-turbo suggest me a list, quote, comment etc. etc


Im trying to accomplish this:

Make gpt-3-5-turbo api output recommendations such as a list(in bullet points), block of quote, an expert comment, a made up review etc. etc. for the text I send to him using embeddings.

So for example:

  • I send 4 blocks of article which is 300 words in total
  • openai api outputs a recommendation of what would be the best to insert at the end of this text (like some list or maybe an anecdote etc.)

How do I accomplish that? I know python and I made a pretty good app that generates a really high quality articles that no human could distinguish, this is the last thing Im missing and I don’t know how to make this happen.

Can any kind soul suggest me how to do this? Ive read about one guy doing this exact thing and he only hinted “embeddings”.

Im already using embeddings in my app to compare some text using cosine similarity function from openai lib for python.

I know im missing 1 puzzle to make this happen and I just can’t figure it out or my brain is overheated lol