My vote is for ids with prefixes. If you've ever used the Stripe API, you know how nice it is that all customer ids are prefixed with `cus_`, all product ids are prefixed with `price_` etc. I want that in postgres, so that any time you see an id you know exactly what kind of entity you are dealing with. It'd be incredibly useful for dealing with data analysts, and also for debugging production issues from logs / crash dumps.

You wouldn't need to store the prefixes at all, it'd be part of the column metadata, but it would be prefixed before sending the data back to the client.

I could do this at the ORM level but in my experience even with the greatest ORM in the world you often have to drop into raw SQL. Additionally people might have multiple ways of accessing the same db or a read-only duplicate.