Can you go into more detail about what you mean by "duplication of data"? I worked on a team whose product uses Hasura; my understanding is that data still comes directly from Postgres, even if Hasura is sitting in the middle. Where is the data being duplicated?