Bram

The evaporation is the evaluation. Getting it in hour one instead of week six is the entire benefit.

The problem is the demo project. Everything is pleasant on a fresh empty project, because the demo was designed by people who wanted it to be, and because none of your actual problems exist there.

So the single change that fixes most of this: evaluate on a real task from your real work. A small one, ideally something slightly awkward that you have already done the old way so you have a comparison.

What that exposes and a demo cannot:

How it behaves on your existing mess rather than on a clean slate.

What happens when something goes wrong, which is where most tools reveal themselves. The error messages and the debugging story are more important than the happy path and are never in the demo.

The second and third hour, when the novelty is gone and you are just working.

Whether it fits everything else you use, which is usually the actual constraint.

Write down before you start what would make you switch: specifically, one or two things. Then you have a criterion rather than a feeling, and if the answer is no you can move on cleanly instead of half-adopting.

Half-adopting is the failure mode to watch for: two of everything is worse than either one.