LLM / GPT integration

[joepio]

You've probably already seen some examples of how integrating a product with a large language model like GPT-3 can create magic appliances.

I'd like to integrate that with Atomic as well.

• Write a plugin description (in swagger / openAPI spec) so GPT can query / use Atomic Server. Should probably limit this to only full-text queries
• Host a manifest file on Atomic Server. See description.
• Write a front-end component that lets the user ask questions. This will probably look like a chat room.