I’m wondering if it is possible and practical to support a document management (search) use case with OpenAI APIs.?
Most document management systems create a massive reverse index for search. Due to the limited context, it seems like the search scope would also be limited.
Do embeddings and vector database somehow help with this use case?
ChatGPT is the web product of Open AI. It’s distinct from the API.
Can you use the Open AI Embeddings API to help you manage retrieval in a document management system? For sure!
Thanks. I updated my post to mention the Open AI APIs. instead of ChatGPT.
Also, do I understand correctly that via the embedding API I might find the most relevant document(s) matching a prompt, and that/those can be added to the context for the prompt? However, there would still be context limitations that could affect the output relevancy for a large document.