Soft question: Erlang/Elixir shell, Rust high performance 'core'?

If we take the Erlang/Elixir philosophy of distributed systems, at some point it states: you can always rewrite the high performance parts in C / C++ / Rust via FFI / ports / tcp+udp.

For those who have hacked on distributed systems where the high performance parts were written in Rust, has it ever been useful to add an Elixir / Erlang 'shell' ? Phrase another way: is there any distributed systems project that, after having done the heavyweight high performance code in Rust, decided to add an Elixir / Erlang 'shell' on top ?