Instead of creating embeddings inline in our event processor using OpenAI API and incurring charges, going to write the message and relevant information to the NATS queue and let a consumer handle vectorizing and storing it back into neo4j vector index.
Instead of creating embeddings inline in our event processor using OpenAI API and incurring charges, going to write the message and relevant information to the NATS queue and let a consumer handle vectorizing and storing it back into neo4j vector index.