Add embeddings to a workflow

[tiim_n8n]

Hey there,

I am building a RAG System. But i am having trouble adding the embeddings to my DB.

I already splitted my text into chunks, so i dont want to use a text splitter therefore i just want to match the already parsed and extracted data from previous nodes into my database schema.

But there is no single node which allows you to get embeddings from text and put that into your json, from what I have seen so far.

The PGVector node doesnt seem to fit for my case since i cannot match my already parsed data and i wrote my own text splitter.

I would be greatful for suggestions.

[Jim_Le]

Hey @tiim_n8n
Hmm seems like it might be easier to just use the built-in character text splitting in “custom” mode and set the character length to a very high amount (eg. 1000000). I think this would ensure the docs stay as one piece.

If you still want to generate embeddings separately in your template, you’ll need to use the HTTP node and request from the embedding provider’s API directly. see https://platform.openai.com/docs/guides/embeddings?lang=curl