Hybrid search-indexer

Hi I want to know this-
I am not using vectorized integration because its still in preview mode. So without this I want to build RAG for my client.
I have below query

1. Indexer can Pull data from ADLS , but in case I want to perform hybrid search which requires embedding this will not work. Until and unless I put json which has both text + Embedding in my ADLS
2. my source data is in ADLS . So now I need to write separate code for doing chunk+embedding using Azure openAI model (lets say I am using Databricks notebook)
3. Whenever any new data comes up in ADLS , I need to call this databricks notebook . So which one will you suggest using Azure function or ADF which will call notebook and create chunks and index it in AI search?
4. is there any other better approach ?can we have custom web api skill in indexer to call embedding model endpoint?is that possible

No repository file, test, or entry point is named. Start by locating samples covering ADLS indexers, embeddings, custom web API skills, and notebook orchestration; done would require a clearly agreed implementation path and corresponding sample or documentation.

Written by the indexing model from the issue text.