I’ve increasingly been using pgvector. It’s an excellent workflow when you can just include similarity search as another where clause in your sql query.