I am studying the Contentful CMS at work. I am also learning chatgpt on my own. I want to create a public-facing chatbot (if that is the correct term) that accepts multiple Contentful-related text files and stores them in a vector db. I want to use Chatgpt to provide inference and answers to questions from a front-end. I am currently using privateGPT successfully to ingest the docs into ChromaDB and gpt4all to query it. This is good for learning but painfully slow and I want to externalize the db to a commercial SAAS product and use Chatgpt3 or 4 as the engine. This will enable me to do development and testing locally and pay to get better and faster results. I’m sure this setup has been done in the past. I’m an experienced Java/SpringBoot developer so I mainly want to get the infrastrutcture in place for this and take it from there. Thanks!
There was no actual question in that, but if your question is “how” to do it, I’d first learn to call the OpenAI HTTP API. Someone on this forum already wrote a Java library for wrapping the HTTP API nicely into Java objects, interfaces. You can find it by searching the forum probably.
Then use this embedding technique described in this video, to search documents in whose embedding you’ll keep in some kind of Vector DB.