Comment by naveedjanmo

Vectorisation is done via OpenAI's embedding API. And the chunking/querying is happens through the Langchain library. But there are a few different ways of doing it - another good library is LLamaIndex.