There are a couple of potentially relevant areas for us:

Firstly the completions API, which can be used for e.g. question answering, and also the embeddings API which effectively provides contextual search and similarity across large volumes of data. The two can be used together, so the embeddings API returns the most relevant section of text pertaining to a question, then the completions API extracts the relevant data and returns it to the user.

There’s an example, using OpenAI’s pre-prepared dataset on Olympics facts in their Cookbook here: https://github.com/openai/openai-cookbook/blob/main/examples/Question_answering_using_embeddings.ipynb