Ephemeral Testing

We have many ways to justify application quality. Unit testing. Fuzz testing. Integration testing. And so forth.

I’d akin to propose a method that was incredible before: ephemeral testing. (Ephemeral is a fancy term for ‘throw away’ or ‘temporary’.)

You compose your code. You build your application component. Or the AI delegate does it for you, it does not matter.

Then you ask an AI delegate to build on it: an application, another layer, perchance several. You have it test what it built. You do not measure the first activity directly. You measure how fine the application built on top of it is.

It is a form of integration testing. The difference is that the application on top is entirely ephemeral. You throw it distant whenever you are done.

A archive alongside a spotless API, stable invariants, and helpful errors lets the delegate create item that plant quickly. A archive alongside hidden state, amazing defaults, or incomplete docs produces a heap of patches and failures. The failures are evidence concerning your code, not concerning the agent.

You can reiterate it. Different agents, distinct tasks, identical foundation.

In effect, alternatively of construction the center during trying to anticipate what power be needed at the another layers, you fair simulate the another layers by really construction them.

Of course, you could contend that alongside AI, you can rebuild everything whenever you need to. But that’s not practical. You need several form of stability.

I have been applying this trick to assorted projects. As I regard a new feature, I ask my AI to prototype quickly what I power afterward build according to what I am doing it. Ephemeral evaluation plant for me thus far.