Summary
In this chapter, we discussed a vector search architecture. You now understand what role it plays in generative AI applications, especially in RAG, and you can deal with core LangChain interfaces for embedding models and vector stores.
We also learned how to use three different managed vector stores on Google Cloud – Vertex Vector Search, pg_vectors with Cloud SQL, and VectorSearch with Big Query. We discussed how to provision the corresponding Google Cloud services, how to ingest data there, and how to retrieve your results based on queries. We also looked into how to retrieve semantically similar texts with pre-filtering based on text metadata.
