Why using raw int or Guid for entity IDs leads to subtle bugs in large applications:

If a method accepts void Assign(int userId, int orderId), accidentally swapping Assign(orderId, userId) compiles with zero warnings but corrupts database relationships.

Strongly Typed IDs:

public readonly record struct UserId(Guid Value);
public readonly record struct OrderId(Guid Value);

public void Assign(UserId user, OrderId order) { /* Type-Safe! */ }

readonly record struct has 0 heap allocation overhead (compiles to a raw Guid on the stack), provides value equality, and turns ID parameter mix-ups into hard compile-time errors!

Strongly typed IDs prevent entire classes of domain model bugs. Works seamlessly with EF Core Value Converters as well.

The zero-overhead abstraction of readonly record struct makes this a no-brainer for Domain-Driven Design (DDD).