To me, creating an API before there are client apps that use the API is putting the cart before the horse. It’s a way to build unnecessary complexity by writing code that handles eventualities that never come up. Start with the client apps, then build exactly enough API for those apps to work. Expand the API incrementally if needed by new apps. That way of working dramatically reduces the odds of an API change.

Source: x.com/allenholub/status/1824167847024791806