Ask HN: What are the best practises to make GPT API to digest different data?

Hi people!

I am a novice regarding AI and I am currently looking for a way to analyse the content of web links, pdfs, slides or youtube videos using the GPT api to have it interact with users.

What are the current best practises regarding this? Should i use embeddings or use some libraries that can extract text from those different data sources?

Thanks for you time!