@IsaacT , I think you are talking about a vectorstore conversational retrieval chain. Meaning that you want to a chatbot with an AI that has access to any kind of knowledge provided.
If that's the case:
I would not recommend building that in Xano, 100% use Flowise or Langflow for that. You need to upsert the docs you want as knowledge on a vector store, which is similar to a database that stores vectors, when you upsert you are splitting text chunks into vectors stored there. The best one is probably Supabase's extension pgvector, but any could work. Once done that, you need to set up the chain in Flowise and connect it with your toddle project as you would do with any chat completions api. If you want the api streamed you'll need to do it on a custom action, for now at least.
If you need each user to upload its own data, you need to create another workflow that upserts their own data with metadata filters so you can filter in each request
I've played with that quite a lot, apart from the quality of data provided if you combine it with prompt chainning it's really good for many use cases
@IsaacT
.jpeg/public)